feat(cli): poll for updates every 10 minutes in serve

Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com>

packages/cli/src/commands/handlers/serve.ts

@@ -1,7 +1,7 @@
 import { NodeHttpServer } from "@effect/platform-node"
 import { Credential } from "@opencode-ai/core/credential"
 import { PermissionSaved } from "@opencode-ai/core/permission/saved"
-import { Context, Layer, Option } from "effect"
+import { Context, Layer, Option, Schedule } from "effect"
 import * as Effect from "effect/Effect"
 import { HttpRouter, HttpServer } from "effect/unstable/http"
 import { createServer } from "node:http"
@@ -34,7 +34,7 @@ export default Runtime.handler(
         const url = HttpServer.formatAddress(address)
         console.log(input.stdio ? JSON.stringify({ url }) : `server listening on ${url}`)
         const updater = yield* Updater.Service
-        yield* updater.check().pipe(Effect.forkScoped)
+        yield* updater.check().pipe(Effect.schedule(Schedule.spaced("10 minutes")), Effect.forkScoped)
         return yield* (input.stdio ? waitForStdinClose() : Effect.never)
       }).pipe(Effect.annotateLogs({ role: "server" })),
     )

packages/client/src/contract.ts

@@ -23,6 +23,7 @@ export const groupNames = {
   "server.session": "sessions",
   "server.message": "messages",
   "server.model": "models",
+  "server.generate": "generate",
   "server.provider": "providers",
   "server.integration": "integrations",
   "server.credential": "credentials",

packages/client/src/generated-effect/client.ts

@@ -260,169 +260,186 @@ const Endpoint5_0 = (raw: RawClient["server.model"]) => (input?: Endpoint5_0Inpu
 
 const adaptGroup5 = (raw: RawClient["server.model"]) => ({ list: Endpoint5_0(raw) })
 
-type Endpoint6_0Request = Parameters<RawClient["server.provider"]["provider.list"]>[0]
-type Endpoint6_0Input = { readonly location?: Endpoint6_0Request["query"]["location"] }
-const Endpoint6_0 = (raw: RawClient["server.provider"]) => (input?: Endpoint6_0Input) =>
+type Endpoint6_0Request = Parameters<RawClient["server.generate"]["generate.text"]>[0]
+type Endpoint6_0Input = {
+  readonly location?: Endpoint6_0Request["query"]["location"]
+  readonly prompt: Endpoint6_0Request["payload"]["prompt"]
+  readonly model?: Endpoint6_0Request["payload"]["model"]
+}
+const Endpoint6_0 = (raw: RawClient["server.generate"]) => (input: Endpoint6_0Input) =>
+  raw["generate.text"]({
+    query: { location: input["location"] },
+    payload: { prompt: input["prompt"], model: input["model"] },
+  }).pipe(
+    Effect.mapError(mapClientError),
+    Effect.map((value) => value.data),
+  )
+
+const adaptGroup6 = (raw: RawClient["server.generate"]) => ({ text: Endpoint6_0(raw) })
+
+type Endpoint7_0Request = Parameters<RawClient["server.provider"]["provider.list"]>[0]
+type Endpoint7_0Input = { readonly location?: Endpoint7_0Request["query"]["location"] }
+const Endpoint7_0 = (raw: RawClient["server.provider"]) => (input?: Endpoint7_0Input) =>
   raw["provider.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint6_1Request = Parameters<RawClient["server.provider"]["provider.get"]>[0]
-type Endpoint6_1Input = {
-  readonly providerID: Endpoint6_1Request["params"]["providerID"]
-  readonly location?: Endpoint6_1Request["query"]["location"]
+type Endpoint7_1Request = Parameters<RawClient["server.provider"]["provider.get"]>[0]
+type Endpoint7_1Input = {
+  readonly providerID: Endpoint7_1Request["params"]["providerID"]
+  readonly location?: Endpoint7_1Request["query"]["location"]
 }
-const Endpoint6_1 = (raw: RawClient["server.provider"]) => (input: Endpoint6_1Input) =>
+const Endpoint7_1 = (raw: RawClient["server.provider"]) => (input: Endpoint7_1Input) =>
   raw["provider.get"]({ params: { providerID: input["providerID"] }, query: { location: input["location"] } }).pipe(
     Effect.mapError(mapClientError),
   )
 
-const adaptGroup6 = (raw: RawClient["server.provider"]) => ({ list: Endpoint6_0(raw), get: Endpoint6_1(raw) })
+const adaptGroup7 = (raw: RawClient["server.provider"]) => ({ list: Endpoint7_0(raw), get: Endpoint7_1(raw) })
 
-type Endpoint7_0Request = Parameters<RawClient["server.integration"]["integration.list"]>[0]
-type Endpoint7_0Input = { readonly location?: Endpoint7_0Request["query"]["location"] }
-const Endpoint7_0 = (raw: RawClient["server.integration"]) => (input?: Endpoint7_0Input) =>
+type Endpoint8_0Request = Parameters<RawClient["server.integration"]["integration.list"]>[0]
+type Endpoint8_0Input = { readonly location?: Endpoint8_0Request["query"]["location"] }
+const Endpoint8_0 = (raw: RawClient["server.integration"]) => (input?: Endpoint8_0Input) =>
   raw["integration.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_1Request = Parameters<RawClient["server.integration"]["integration.get"]>[0]
-type Endpoint7_1Input = {
-  readonly integrationID: Endpoint7_1Request["params"]["integrationID"]
-  readonly location?: Endpoint7_1Request["query"]["location"]
+type Endpoint8_1Request = Parameters<RawClient["server.integration"]["integration.get"]>[0]
+type Endpoint8_1Input = {
+  readonly integrationID: Endpoint8_1Request["params"]["integrationID"]
+  readonly location?: Endpoint8_1Request["query"]["location"]
 }
-const Endpoint7_1 = (raw: RawClient["server.integration"]) => (input: Endpoint7_1Input) =>
+const Endpoint8_1 = (raw: RawClient["server.integration"]) => (input: Endpoint8_1Input) =>
   raw["integration.get"]({
     params: { integrationID: input["integrationID"] },
     query: { location: input["location"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_2Request = Parameters<RawClient["server.integration"]["integration.connect.key"]>[0]
-type Endpoint7_2Input = {
-  readonly integrationID: Endpoint7_2Request["params"]["integrationID"]
-  readonly location?: Endpoint7_2Request["query"]["location"]
-  readonly key: Endpoint7_2Request["payload"]["key"]
-  readonly label?: Endpoint7_2Request["payload"]["label"]
+type Endpoint8_2Request = Parameters<RawClient["server.integration"]["integration.connect.key"]>[0]
+type Endpoint8_2Input = {
+  readonly integrationID: Endpoint8_2Request["params"]["integrationID"]
+  readonly location?: Endpoint8_2Request["query"]["location"]
+  readonly key: Endpoint8_2Request["payload"]["key"]
+  readonly label?: Endpoint8_2Request["payload"]["label"]
 }
-const Endpoint7_2 = (raw: RawClient["server.integration"]) => (input: Endpoint7_2Input) =>
+const Endpoint8_2 = (raw: RawClient["server.integration"]) => (input: Endpoint8_2Input) =>
   raw["integration.connect.key"]({
     params: { integrationID: input["integrationID"] },
     query: { location: input["location"] },
     payload: { key: input["key"], label: input["label"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_3Request = Parameters<RawClient["server.integration"]["integration.connect.oauth"]>[0]
-type Endpoint7_3Input = {
-  readonly integrationID: Endpoint7_3Request["params"]["integrationID"]
-  readonly location?: Endpoint7_3Request["query"]["location"]
-  readonly methodID: Endpoint7_3Request["payload"]["methodID"]
-  readonly inputs: Endpoint7_3Request["payload"]["inputs"]
-  readonly label?: Endpoint7_3Request["payload"]["label"]
+type Endpoint8_3Request = Parameters<RawClient["server.integration"]["integration.connect.oauth"]>[0]
+type Endpoint8_3Input = {
+  readonly integrationID: Endpoint8_3Request["params"]["integrationID"]
+  readonly location?: Endpoint8_3Request["query"]["location"]
+  readonly methodID: Endpoint8_3Request["payload"]["methodID"]
+  readonly inputs: Endpoint8_3Request["payload"]["inputs"]
+  readonly label?: Endpoint8_3Request["payload"]["label"]
 }
-const Endpoint7_3 = (raw: RawClient["server.integration"]) => (input: Endpoint7_3Input) =>
+const Endpoint8_3 = (raw: RawClient["server.integration"]) => (input: Endpoint8_3Input) =>
   raw["integration.connect.oauth"]({
     params: { integrationID: input["integrationID"] },
     query: { location: input["location"] },
     payload: { methodID: input["methodID"], inputs: input["inputs"], label: input["label"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_4Request = Parameters<RawClient["server.integration"]["integration.attempt.status"]>[0]
-type Endpoint7_4Input = {
-  readonly attemptID: Endpoint7_4Request["params"]["attemptID"]
-  readonly location?: Endpoint7_4Request["query"]["location"]
+type Endpoint8_4Request = Parameters<RawClient["server.integration"]["integration.attempt.status"]>[0]
+type Endpoint8_4Input = {
+  readonly attemptID: Endpoint8_4Request["params"]["attemptID"]
+  readonly location?: Endpoint8_4Request["query"]["location"]
 }
-const Endpoint7_4 = (raw: RawClient["server.integration"]) => (input: Endpoint7_4Input) =>
+const Endpoint8_4 = (raw: RawClient["server.integration"]) => (input: Endpoint8_4Input) =>
   raw["integration.attempt.status"]({
     params: { attemptID: input["attemptID"] },
     query: { location: input["location"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_5Request = Parameters<RawClient["server.integration"]["integration.attempt.complete"]>[0]
-type Endpoint7_5Input = {
-  readonly attemptID: Endpoint7_5Request["params"]["attemptID"]
-  readonly location?: Endpoint7_5Request["query"]["location"]
-  readonly code?: Endpoint7_5Request["payload"]["code"]
+type Endpoint8_5Request = Parameters<RawClient["server.integration"]["integration.attempt.complete"]>[0]
+type Endpoint8_5Input = {
+  readonly attemptID: Endpoint8_5Request["params"]["attemptID"]
+  readonly location?: Endpoint8_5Request["query"]["location"]
+  readonly code?: Endpoint8_5Request["payload"]["code"]
 }
-const Endpoint7_5 = (raw: RawClient["server.integration"]) => (input: Endpoint7_5Input) =>
+const Endpoint8_5 = (raw: RawClient["server.integration"]) => (input: Endpoint8_5Input) =>
   raw["integration.attempt.complete"]({
     params: { attemptID: input["attemptID"] },
     query: { location: input["location"] },
     payload: { code: input["code"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint7_6Request = Parameters<RawClient["server.integration"]["integration.attempt.cancel"]>[0]
-type Endpoint7_6Input = {
-  readonly attemptID: Endpoint7_6Request["params"]["attemptID"]
-  readonly location?: Endpoint7_6Request["query"]["location"]
+type Endpoint8_6Request = Parameters<RawClient["server.integration"]["integration.attempt.cancel"]>[0]
+type Endpoint8_6Input = {
+  readonly attemptID: Endpoint8_6Request["params"]["attemptID"]
+  readonly location?: Endpoint8_6Request["query"]["location"]
 }
-const Endpoint7_6 = (raw: RawClient["server.integration"]) => (input: Endpoint7_6Input) =>
+const Endpoint8_6 = (raw: RawClient["server.integration"]) => (input: Endpoint8_6Input) =>
   raw["integration.attempt.cancel"]({
     params: { attemptID: input["attemptID"] },
     query: { location: input["location"] },
   }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup7 = (raw: RawClient["server.integration"]) => ({
-  list: Endpoint7_0(raw),
-  get: Endpoint7_1(raw),
-  connectKey: Endpoint7_2(raw),
-  connectOauth: Endpoint7_3(raw),
-  attemptStatus: Endpoint7_4(raw),
-  attemptComplete: Endpoint7_5(raw),
-  attemptCancel: Endpoint7_6(raw),
+const adaptGroup8 = (raw: RawClient["server.integration"]) => ({
+  list: Endpoint8_0(raw),
+  get: Endpoint8_1(raw),
+  connectKey: Endpoint8_2(raw),
+  connectOauth: Endpoint8_3(raw),
+  attemptStatus: Endpoint8_4(raw),
+  attemptComplete: Endpoint8_5(raw),
+  attemptCancel: Endpoint8_6(raw),
 })
 
-type Endpoint8_0Request = Parameters<RawClient["server.credential"]["credential.update"]>[0]
-type Endpoint8_0Input = {
-  readonly credentialID: Endpoint8_0Request["params"]["credentialID"]
-  readonly location?: Endpoint8_0Request["query"]["location"]
-  readonly label: Endpoint8_0Request["payload"]["label"]
+type Endpoint9_0Request = Parameters<RawClient["server.credential"]["credential.update"]>[0]
+type Endpoint9_0Input = {
+  readonly credentialID: Endpoint9_0Request["params"]["credentialID"]
+  readonly location?: Endpoint9_0Request["query"]["location"]
+  readonly label: Endpoint9_0Request["payload"]["label"]
 }
-const Endpoint8_0 = (raw: RawClient["server.credential"]) => (input: Endpoint8_0Input) =>
+const Endpoint9_0 = (raw: RawClient["server.credential"]) => (input: Endpoint9_0Input) =>
   raw["credential.update"]({
     params: { credentialID: input["credentialID"] },
     query: { location: input["location"] },
     payload: { label: input["label"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint8_1Request = Parameters<RawClient["server.credential"]["credential.remove"]>[0]
-type Endpoint8_1Input = {
-  readonly credentialID: Endpoint8_1Request["params"]["credentialID"]
-  readonly location?: Endpoint8_1Request["query"]["location"]
+type Endpoint9_1Request = Parameters<RawClient["server.credential"]["credential.remove"]>[0]
+type Endpoint9_1Input = {
+  readonly credentialID: Endpoint9_1Request["params"]["credentialID"]
+  readonly location?: Endpoint9_1Request["query"]["location"]
 }
-const Endpoint8_1 = (raw: RawClient["server.credential"]) => (input: Endpoint8_1Input) =>
+const Endpoint9_1 = (raw: RawClient["server.credential"]) => (input: Endpoint9_1Input) =>
   raw["credential.remove"]({
     params: { credentialID: input["credentialID"] },
     query: { location: input["location"] },
   }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup8 = (raw: RawClient["server.credential"]) => ({ update: Endpoint8_0(raw), remove: Endpoint8_1(raw) })
+const adaptGroup9 = (raw: RawClient["server.credential"]) => ({ update: Endpoint9_0(raw), remove: Endpoint9_1(raw) })
 
-type Endpoint9_0Request = Parameters<RawClient["server.permission"]["permission.request.list"]>[0]
-type Endpoint9_0Input = { readonly location?: Endpoint9_0Request["query"]["location"] }
-const Endpoint9_0 = (raw: RawClient["server.permission"]) => (input?: Endpoint9_0Input) =>
+type Endpoint10_0Request = Parameters<RawClient["server.permission"]["permission.request.list"]>[0]
+type Endpoint10_0Input = { readonly location?: Endpoint10_0Request["query"]["location"] }
+const Endpoint10_0 = (raw: RawClient["server.permission"]) => (input?: Endpoint10_0Input) =>
   raw["permission.request.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint9_1Request = Parameters<RawClient["server.permission"]["permission.saved.list"]>[0]
-type Endpoint9_1Input = { readonly projectID?: Endpoint9_1Request["query"]["projectID"] }
-const Endpoint9_1 = (raw: RawClient["server.permission"]) => (input?: Endpoint9_1Input) =>
+type Endpoint10_1Request = Parameters<RawClient["server.permission"]["permission.saved.list"]>[0]
+type Endpoint10_1Input = { readonly projectID?: Endpoint10_1Request["query"]["projectID"] }
+const Endpoint10_1 = (raw: RawClient["server.permission"]) => (input?: Endpoint10_1Input) =>
   raw["permission.saved.list"]({ query: { projectID: input?.["projectID"] } }).pipe(
     Effect.mapError(mapClientError),
     Effect.map((value) => value.data),
   )
 
-type Endpoint9_2Request = Parameters<RawClient["server.permission"]["permission.saved.remove"]>[0]
-type Endpoint9_2Input = { readonly id: Endpoint9_2Request["params"]["id"] }
-const Endpoint9_2 = (raw: RawClient["server.permission"]) => (input: Endpoint9_2Input) =>
+type Endpoint10_2Request = Parameters<RawClient["server.permission"]["permission.saved.remove"]>[0]
+type Endpoint10_2Input = { readonly id: Endpoint10_2Request["params"]["id"] }
+const Endpoint10_2 = (raw: RawClient["server.permission"]) => (input: Endpoint10_2Input) =>
   raw["permission.saved.remove"]({ params: { id: input["id"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint9_3Request = Parameters<RawClient["server.permission"]["session.permission.create"]>[0]
-type Endpoint9_3Input = {
-  readonly sessionID: Endpoint9_3Request["params"]["sessionID"]
-  readonly id?: Endpoint9_3Request["payload"]["id"]
-  readonly action: Endpoint9_3Request["payload"]["action"]
-  readonly resources: Endpoint9_3Request["payload"]["resources"]
-  readonly save?: Endpoint9_3Request["payload"]["save"]
-  readonly metadata?: Endpoint9_3Request["payload"]["metadata"]
-  readonly source?: Endpoint9_3Request["payload"]["source"]
-  readonly agent?: Endpoint9_3Request["payload"]["agent"]
+type Endpoint10_3Request = Parameters<RawClient["server.permission"]["session.permission.create"]>[0]
+type Endpoint10_3Input = {
+  readonly sessionID: Endpoint10_3Request["params"]["sessionID"]
+  readonly id?: Endpoint10_3Request["payload"]["id"]
+  readonly action: Endpoint10_3Request["payload"]["action"]
+  readonly resources: Endpoint10_3Request["payload"]["resources"]
+  readonly save?: Endpoint10_3Request["payload"]["save"]
+  readonly metadata?: Endpoint10_3Request["payload"]["metadata"]
+  readonly source?: Endpoint10_3Request["payload"]["source"]
+  readonly agent?: Endpoint10_3Request["payload"]["agent"]
 }
-const Endpoint9_3 = (raw: RawClient["server.permission"]) => (input: Endpoint9_3Input) =>
+const Endpoint10_3 = (raw: RawClient["server.permission"]) => (input: Endpoint10_3Input) =>
   raw["session.permission.create"]({
     params: { sessionID: input["sessionID"] },
     payload: {
@@ -439,87 +456,87 @@ const Endpoint9_3 = (raw: RawClient["server.permission"]) => (input: Endpoint9_3
     Effect.map((value) => value.data),
   )
 
-type Endpoint9_4Request = Parameters<RawClient["server.permission"]["session.permission.list"]>[0]
-type Endpoint9_4Input = { readonly sessionID: Endpoint9_4Request["params"]["sessionID"] }
-const Endpoint9_4 = (raw: RawClient["server.permission"]) => (input: Endpoint9_4Input) =>
+type Endpoint10_4Request = Parameters<RawClient["server.permission"]["session.permission.list"]>[0]
+type Endpoint10_4Input = { readonly sessionID: Endpoint10_4Request["params"]["sessionID"] }
+const Endpoint10_4 = (raw: RawClient["server.permission"]) => (input: Endpoint10_4Input) =>
   raw["session.permission.list"]({ params: { sessionID: input["sessionID"] } }).pipe(
     Effect.mapError(mapClientError),
     Effect.map((value) => value.data),
   )
 
-type Endpoint9_5Request = Parameters<RawClient["server.permission"]["session.permission.get"]>[0]
-type Endpoint9_5Input = {
-  readonly sessionID: Endpoint9_5Request["params"]["sessionID"]
-  readonly requestID: Endpoint9_5Request["params"]["requestID"]
+type Endpoint10_5Request = Parameters<RawClient["server.permission"]["session.permission.get"]>[0]
+type Endpoint10_5Input = {
+  readonly sessionID: Endpoint10_5Request["params"]["sessionID"]
+  readonly requestID: Endpoint10_5Request["params"]["requestID"]
 }
-const Endpoint9_5 = (raw: RawClient["server.permission"]) => (input: Endpoint9_5Input) =>
+const Endpoint10_5 = (raw: RawClient["server.permission"]) => (input: Endpoint10_5Input) =>
   raw["session.permission.get"]({ params: { sessionID: input["sessionID"], requestID: input["requestID"] } }).pipe(
     Effect.mapError(mapClientError),
     Effect.map((value) => value.data),
   )
 
-type Endpoint9_6Request = Parameters<RawClient["server.permission"]["session.permission.reply"]>[0]
-type Endpoint9_6Input = {
-  readonly sessionID: Endpoint9_6Request["params"]["sessionID"]
-  readonly requestID: Endpoint9_6Request["params"]["requestID"]
-  readonly reply: Endpoint9_6Request["payload"]["reply"]
-  readonly message?: Endpoint9_6Request["payload"]["message"]
+type Endpoint10_6Request = Parameters<RawClient["server.permission"]["session.permission.reply"]>[0]
+type Endpoint10_6Input = {
+  readonly sessionID: Endpoint10_6Request["params"]["sessionID"]
+  readonly requestID: Endpoint10_6Request["params"]["requestID"]
+  readonly reply: Endpoint10_6Request["payload"]["reply"]
+  readonly message?: Endpoint10_6Request["payload"]["message"]
 }
-const Endpoint9_6 = (raw: RawClient["server.permission"]) => (input: Endpoint9_6Input) =>
+const Endpoint10_6 = (raw: RawClient["server.permission"]) => (input: Endpoint10_6Input) =>
   raw["session.permission.reply"]({
     params: { sessionID: input["sessionID"], requestID: input["requestID"] },
     payload: { reply: input["reply"], message: input["message"] },
   }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup9 = (raw: RawClient["server.permission"]) => ({
-  listRequests: Endpoint9_0(raw),
-  listSaved: Endpoint9_1(raw),
-  removeSaved: Endpoint9_2(raw),
-  create: Endpoint9_3(raw),
-  list: Endpoint9_4(raw),
-  get: Endpoint9_5(raw),
-  reply: Endpoint9_6(raw),
+const adaptGroup10 = (raw: RawClient["server.permission"]) => ({
+  listRequests: Endpoint10_0(raw),
+  listSaved: Endpoint10_1(raw),
+  removeSaved: Endpoint10_2(raw),
+  create: Endpoint10_3(raw),
+  list: Endpoint10_4(raw),
+  get: Endpoint10_5(raw),
+  reply: Endpoint10_6(raw),
 })
 
-type Endpoint10_0Request = Parameters<RawClient["server.fs"]["fs.list"]>[0]
-type Endpoint10_0Input = {
-  readonly location?: Endpoint10_0Request["query"]["location"]
-  readonly path?: Endpoint10_0Request["query"]["path"]
+type Endpoint11_0Request = Parameters<RawClient["server.fs"]["fs.list"]>[0]
+type Endpoint11_0Input = {
+  readonly location?: Endpoint11_0Request["query"]["location"]
+  readonly path?: Endpoint11_0Request["query"]["path"]
 }
-const Endpoint10_0 = (raw: RawClient["server.fs"]) => (input?: Endpoint10_0Input) =>
+const Endpoint11_0 = (raw: RawClient["server.fs"]) => (input?: Endpoint11_0Input) =>
   raw["fs.list"]({ query: { location: input?.["location"], path: input?.["path"] } }).pipe(
     Effect.mapError(mapClientError),
   )
 
-type Endpoint10_1Request = Parameters<RawClient["server.fs"]["fs.find"]>[0]
-type Endpoint10_1Input = {
-  readonly location?: Endpoint10_1Request["query"]["location"]
-  readonly query: Endpoint10_1Request["query"]["query"]
-  readonly type?: Endpoint10_1Request["query"]["type"]
-  readonly limit?: Endpoint10_1Request["query"]["limit"]
+type Endpoint11_1Request = Parameters<RawClient["server.fs"]["fs.find"]>[0]
+type Endpoint11_1Input = {
+  readonly location?: Endpoint11_1Request["query"]["location"]
+  readonly query: Endpoint11_1Request["query"]["query"]
+  readonly type?: Endpoint11_1Request["query"]["type"]
+  readonly limit?: Endpoint11_1Request["query"]["limit"]
 }
-const Endpoint10_1 = (raw: RawClient["server.fs"]) => (input: Endpoint10_1Input) =>
+const Endpoint11_1 = (raw: RawClient["server.fs"]) => (input: Endpoint11_1Input) =>
   raw["fs.find"]({
     query: { location: input["location"], query: input["query"], type: input["type"], limit: input["limit"] },
   }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup10 = (raw: RawClient["server.fs"]) => ({ list: Endpoint10_0(raw), find: Endpoint10_1(raw) })
+const adaptGroup11 = (raw: RawClient["server.fs"]) => ({ list: Endpoint11_0(raw), find: Endpoint11_1(raw) })
 
-type Endpoint11_0Request = Parameters<RawClient["server.command"]["command.list"]>[0]
-type Endpoint11_0Input = { readonly location?: Endpoint11_0Request["query"]["location"] }
-const Endpoint11_0 = (raw: RawClient["server.command"]) => (input?: Endpoint11_0Input) =>
+type Endpoint12_0Request = Parameters<RawClient["server.command"]["command.list"]>[0]
+type Endpoint12_0Input = { readonly location?: Endpoint12_0Request["query"]["location"] }
+const Endpoint12_0 = (raw: RawClient["server.command"]) => (input?: Endpoint12_0Input) =>
   raw["command.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup11 = (raw: RawClient["server.command"]) => ({ list: Endpoint11_0(raw) })
+const adaptGroup12 = (raw: RawClient["server.command"]) => ({ list: Endpoint12_0(raw) })
 
-type Endpoint12_0Request = Parameters<RawClient["server.skill"]["skill.list"]>[0]
-type Endpoint12_0Input = { readonly location?: Endpoint12_0Request["query"]["location"] }
-const Endpoint12_0 = (raw: RawClient["server.skill"]) => (input?: Endpoint12_0Input) =>
+type Endpoint13_0Request = Parameters<RawClient["server.skill"]["skill.list"]>[0]
+type Endpoint13_0Input = { readonly location?: Endpoint13_0Request["query"]["location"] }
+const Endpoint13_0 = (raw: RawClient["server.skill"]) => (input?: Endpoint13_0Input) =>
   raw["skill.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup12 = (raw: RawClient["server.skill"]) => ({ list: Endpoint12_0(raw) })
+const adaptGroup13 = (raw: RawClient["server.skill"]) => ({ list: Endpoint13_0(raw) })
 
-const Endpoint13_0 = (raw: RawClient["server.event"]) => () =>
+const Endpoint14_0 = (raw: RawClient["server.event"]) => () =>
   Stream.unwrap(
     raw["event.subscribe"]({}).pipe(
       Effect.mapError(mapClientError),
@@ -527,23 +544,23 @@ const Endpoint13_0 = (raw: RawClient["server.event"]) => () =>
     ),
   )
 
-const adaptGroup13 = (raw: RawClient["server.event"]) => ({ subscribe: Endpoint13_0(raw) })
+const adaptGroup14 = (raw: RawClient["server.event"]) => ({ subscribe: Endpoint14_0(raw) })
 
-type Endpoint14_0Request = Parameters<RawClient["server.pty"]["pty.list"]>[0]
-type Endpoint14_0Input = { readonly location?: Endpoint14_0Request["query"]["location"] }
-const Endpoint14_0 = (raw: RawClient["server.pty"]) => (input?: Endpoint14_0Input) =>
+type Endpoint15_0Request = Parameters<RawClient["server.pty"]["pty.list"]>[0]
+type Endpoint15_0Input = { readonly location?: Endpoint15_0Request["query"]["location"] }
+const Endpoint15_0 = (raw: RawClient["server.pty"]) => (input?: Endpoint15_0Input) =>
   raw["pty.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint14_1Request = Parameters<RawClient["server.pty"]["pty.create"]>[0]
-type Endpoint14_1Input = {
-  readonly location?: Endpoint14_1Request["query"]["location"]
-  readonly command?: Endpoint14_1Request["payload"]["command"]
-  readonly args?: Endpoint14_1Request["payload"]["args"]
-  readonly cwd?: Endpoint14_1Request["payload"]["cwd"]
-  readonly title?: Endpoint14_1Request["payload"]["title"]
-  readonly env?: Endpoint14_1Request["payload"]["env"]
+type Endpoint15_1Request = Parameters<RawClient["server.pty"]["pty.create"]>[0]
+type Endpoint15_1Input = {
+  readonly location?: Endpoint15_1Request["query"]["location"]
+  readonly command?: Endpoint15_1Request["payload"]["command"]
+  readonly args?: Endpoint15_1Request["payload"]["args"]
+  readonly cwd?: Endpoint15_1Request["payload"]["cwd"]
+  readonly title?: Endpoint15_1Request["payload"]["title"]
+  readonly env?: Endpoint15_1Request["payload"]["env"]
 }
-const Endpoint14_1 = (raw: RawClient["server.pty"]) => (input?: Endpoint14_1Input) =>
+const Endpoint15_1 = (raw: RawClient["server.pty"]) => (input?: Endpoint15_1Input) =>
   raw["pty.create"]({
     query: { location: input?.["location"] },
     payload: {
@@ -555,141 +572,201 @@ const Endpoint14_1 = (raw: RawClient["server.pty"]) => (input?: Endpoint14_1Inpu
     },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint14_2Request = Parameters<RawClient["server.pty"]["pty.get"]>[0]
-type Endpoint14_2Input = {
-  readonly ptyID: Endpoint14_2Request["params"]["ptyID"]
-  readonly location?: Endpoint14_2Request["query"]["location"]
+type Endpoint15_2Request = Parameters<RawClient["server.pty"]["pty.get"]>[0]
+type Endpoint15_2Input = {
+  readonly ptyID: Endpoint15_2Request["params"]["ptyID"]
+  readonly location?: Endpoint15_2Request["query"]["location"]
 }
-const Endpoint14_2 = (raw: RawClient["server.pty"]) => (input: Endpoint14_2Input) =>
+const Endpoint15_2 = (raw: RawClient["server.pty"]) => (input: Endpoint15_2Input) =>
   raw["pty.get"]({ params: { ptyID: input["ptyID"] }, query: { location: input["location"] } }).pipe(
     Effect.mapError(mapClientError),
   )
 
-type Endpoint14_3Request = Parameters<RawClient["server.pty"]["pty.update"]>[0]
-type Endpoint14_3Input = {
-  readonly ptyID: Endpoint14_3Request["params"]["ptyID"]
-  readonly location?: Endpoint14_3Request["query"]["location"]
-  readonly title?: Endpoint14_3Request["payload"]["title"]
-  readonly size?: Endpoint14_3Request["payload"]["size"]
+type Endpoint15_3Request = Parameters<RawClient["server.pty"]["pty.update"]>[0]
+type Endpoint15_3Input = {
+  readonly ptyID: Endpoint15_3Request["params"]["ptyID"]
+  readonly location?: Endpoint15_3Request["query"]["location"]
+  readonly title?: Endpoint15_3Request["payload"]["title"]
+  readonly size?: Endpoint15_3Request["payload"]["size"]
 }
-const Endpoint14_3 = (raw: RawClient["server.pty"]) => (input: Endpoint14_3Input) =>
+const Endpoint15_3 = (raw: RawClient["server.pty"]) => (input: Endpoint15_3Input) =>
   raw["pty.update"]({
     params: { ptyID: input["ptyID"] },
     query: { location: input["location"] },
     payload: { title: input["title"], size: input["size"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint14_4Request = Parameters<RawClient["server.pty"]["pty.remove"]>[0]
-type Endpoint14_4Input = {
-  readonly ptyID: Endpoint14_4Request["params"]["ptyID"]
-  readonly location?: Endpoint14_4Request["query"]["location"]
+type Endpoint15_4Request = Parameters<RawClient["server.pty"]["pty.remove"]>[0]
+type Endpoint15_4Input = {
+  readonly ptyID: Endpoint15_4Request["params"]["ptyID"]
+  readonly location?: Endpoint15_4Request["query"]["location"]
 }
-const Endpoint14_4 = (raw: RawClient["server.pty"]) => (input: Endpoint14_4Input) =>
+const Endpoint15_4 = (raw: RawClient["server.pty"]) => (input: Endpoint15_4Input) =>
   raw["pty.remove"]({ params: { ptyID: input["ptyID"] }, query: { location: input["location"] } }).pipe(
     Effect.mapError(mapClientError),
   )
 
-const adaptGroup14 = (raw: RawClient["server.pty"]) => ({
-  list: Endpoint14_0(raw),
-  create: Endpoint14_1(raw),
-  get: Endpoint14_2(raw),
-  update: Endpoint14_3(raw),
-  remove: Endpoint14_4(raw),
+const adaptGroup15 = (raw: RawClient["server.pty"]) => ({
+  list: Endpoint15_0(raw),
+  create: Endpoint15_1(raw),
+  get: Endpoint15_2(raw),
+  update: Endpoint15_3(raw),
+  remove: Endpoint15_4(raw),
 })
 
-type Endpoint15_0Request = Parameters<RawClient["server.question"]["question.request.list"]>[0]
-type Endpoint15_0Input = { readonly location?: Endpoint15_0Request["query"]["location"] }
-const Endpoint15_0 = (raw: RawClient["server.question"]) => (input?: Endpoint15_0Input) =>
+type Endpoint16_0Request = Parameters<RawClient["server.shell"]["shell.list"]>[0]
+type Endpoint16_0Input = { readonly location?: Endpoint16_0Request["query"]["location"] }
+const Endpoint16_0 = (raw: RawClient["server.shell"]) => (input?: Endpoint16_0Input) =>
+  raw["shell.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
+
+type Endpoint16_1Request = Parameters<RawClient["server.shell"]["shell.create"]>[0]
+type Endpoint16_1Input = {
+  readonly location?: Endpoint16_1Request["query"]["location"]
+  readonly command: Endpoint16_1Request["payload"]["command"]
+  readonly cwd?: Endpoint16_1Request["payload"]["cwd"]
+  readonly timeout?: Endpoint16_1Request["payload"]["timeout"]
+  readonly metadata?: Endpoint16_1Request["payload"]["metadata"]
+}
+const Endpoint16_1 = (raw: RawClient["server.shell"]) => (input: Endpoint16_1Input) =>
+  raw["shell.create"]({
+    query: { location: input["location"] },
+    payload: { command: input["command"], cwd: input["cwd"], timeout: input["timeout"], metadata: input["metadata"] },
+  }).pipe(Effect.mapError(mapClientError))
+
+type Endpoint16_2Request = Parameters<RawClient["server.shell"]["shell.get"]>[0]
+type Endpoint16_2Input = {
+  readonly id: Endpoint16_2Request["params"]["id"]
+  readonly location?: Endpoint16_2Request["query"]["location"]
+}
+const Endpoint16_2 = (raw: RawClient["server.shell"]) => (input: Endpoint16_2Input) =>
+  raw["shell.get"]({ params: { id: input["id"] }, query: { location: input["location"] } }).pipe(
+    Effect.mapError(mapClientError),
+  )
+
+type Endpoint16_3Request = Parameters<RawClient["server.shell"]["shell.output"]>[0]
+type Endpoint16_3Input = {
+  readonly id: Endpoint16_3Request["params"]["id"]
+  readonly location?: Endpoint16_3Request["query"]["location"]
+  readonly cursor?: Endpoint16_3Request["query"]["cursor"]
+  readonly limit?: Endpoint16_3Request["query"]["limit"]
+}
+const Endpoint16_3 = (raw: RawClient["server.shell"]) => (input: Endpoint16_3Input) =>
+  raw["shell.output"]({
+    params: { id: input["id"] },
+    query: { location: input["location"], cursor: input["cursor"], limit: input["limit"] },
+  }).pipe(Effect.mapError(mapClientError))
+
+type Endpoint16_4Request = Parameters<RawClient["server.shell"]["shell.remove"]>[0]
+type Endpoint16_4Input = {
+  readonly id: Endpoint16_4Request["params"]["id"]
+  readonly location?: Endpoint16_4Request["query"]["location"]
+}
+const Endpoint16_4 = (raw: RawClient["server.shell"]) => (input: Endpoint16_4Input) =>
+  raw["shell.remove"]({ params: { id: input["id"] }, query: { location: input["location"] } }).pipe(
+    Effect.mapError(mapClientError),
+  )
+
+const adaptGroup16 = (raw: RawClient["server.shell"]) => ({
+  list: Endpoint16_0(raw),
+  create: Endpoint16_1(raw),
+  get: Endpoint16_2(raw),
+  output: Endpoint16_3(raw),
+  remove: Endpoint16_4(raw),
+})
+
+type Endpoint17_0Request = Parameters<RawClient["server.question"]["question.request.list"]>[0]
+type Endpoint17_0Input = { readonly location?: Endpoint17_0Request["query"]["location"] }
+const Endpoint17_0 = (raw: RawClient["server.question"]) => (input?: Endpoint17_0Input) =>
   raw["question.request.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint15_1Request = Parameters<RawClient["server.question"]["session.question.list"]>[0]
-type Endpoint15_1Input = { readonly sessionID: Endpoint15_1Request["params"]["sessionID"] }
-const Endpoint15_1 = (raw: RawClient["server.question"]) => (input: Endpoint15_1Input) =>
+type Endpoint17_1Request = Parameters<RawClient["server.question"]["session.question.list"]>[0]
+type Endpoint17_1Input = { readonly sessionID: Endpoint17_1Request["params"]["sessionID"] }
+const Endpoint17_1 = (raw: RawClient["server.question"]) => (input: Endpoint17_1Input) =>
   raw["session.question.list"]({ params: { sessionID: input["sessionID"] } }).pipe(
     Effect.mapError(mapClientError),
     Effect.map((value) => value.data),
   )
 
-type Endpoint15_2Request = Parameters<RawClient["server.question"]["session.question.reply"]>[0]
-type Endpoint15_2Input = {
-  readonly sessionID: Endpoint15_2Request["params"]["sessionID"]
-  readonly requestID: Endpoint15_2Request["params"]["requestID"]
-  readonly answers: Endpoint15_2Request["payload"]["answers"]
+type Endpoint17_2Request = Parameters<RawClient["server.question"]["session.question.reply"]>[0]
+type Endpoint17_2Input = {
+  readonly sessionID: Endpoint17_2Request["params"]["sessionID"]
+  readonly requestID: Endpoint17_2Request["params"]["requestID"]
+  readonly answers: Endpoint17_2Request["payload"]["answers"]
 }
-const Endpoint15_2 = (raw: RawClient["server.question"]) => (input: Endpoint15_2Input) =>
+const Endpoint17_2 = (raw: RawClient["server.question"]) => (input: Endpoint17_2Input) =>
   raw["session.question.reply"]({
     params: { sessionID: input["sessionID"], requestID: input["requestID"] },
     payload: { answers: input["answers"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint15_3Request = Parameters<RawClient["server.question"]["session.question.reject"]>[0]
-type Endpoint15_3Input = {
-  readonly sessionID: Endpoint15_3Request["params"]["sessionID"]
-  readonly requestID: Endpoint15_3Request["params"]["requestID"]
+type Endpoint17_3Request = Parameters<RawClient["server.question"]["session.question.reject"]>[0]
+type Endpoint17_3Input = {
+  readonly sessionID: Endpoint17_3Request["params"]["sessionID"]
+  readonly requestID: Endpoint17_3Request["params"]["requestID"]
 }
-const Endpoint15_3 = (raw: RawClient["server.question"]) => (input: Endpoint15_3Input) =>
+const Endpoint17_3 = (raw: RawClient["server.question"]) => (input: Endpoint17_3Input) =>
   raw["session.question.reject"]({ params: { sessionID: input["sessionID"], requestID: input["requestID"] } }).pipe(
     Effect.mapError(mapClientError),
   )
 
-const adaptGroup15 = (raw: RawClient["server.question"]) => ({
-  listRequests: Endpoint15_0(raw),
-  list: Endpoint15_1(raw),
-  reply: Endpoint15_2(raw),
-  reject: Endpoint15_3(raw),
+const adaptGroup17 = (raw: RawClient["server.question"]) => ({
+  listRequests: Endpoint17_0(raw),
+  list: Endpoint17_1(raw),
+  reply: Endpoint17_2(raw),
+  reject: Endpoint17_3(raw),
 })
 
-type Endpoint16_0Request = Parameters<RawClient["server.reference"]["reference.list"]>[0]
-type Endpoint16_0Input = { readonly location?: Endpoint16_0Request["query"]["location"] }
-const Endpoint16_0 = (raw: RawClient["server.reference"]) => (input?: Endpoint16_0Input) =>
+type Endpoint18_0Request = Parameters<RawClient["server.reference"]["reference.list"]>[0]
+type Endpoint18_0Input = { readonly location?: Endpoint18_0Request["query"]["location"] }
+const Endpoint18_0 = (raw: RawClient["server.reference"]) => (input?: Endpoint18_0Input) =>
   raw["reference.list"]({ query: { location: input?.["location"] } }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup16 = (raw: RawClient["server.reference"]) => ({ list: Endpoint16_0(raw) })
+const adaptGroup18 = (raw: RawClient["server.reference"]) => ({ list: Endpoint18_0(raw) })
 
-type Endpoint17_0Request = Parameters<RawClient["server.projectCopy"]["projectCopy.create"]>[0]
-type Endpoint17_0Input = {
-  readonly projectID: Endpoint17_0Request["params"]["projectID"]
-  readonly location?: Endpoint17_0Request["query"]["location"]
-  readonly strategy: Endpoint17_0Request["payload"]["strategy"]
-  readonly directory: Endpoint17_0Request["payload"]["directory"]
-  readonly name?: Endpoint17_0Request["payload"]["name"]
+type Endpoint19_0Request = Parameters<RawClient["server.projectCopy"]["projectCopy.create"]>[0]
+type Endpoint19_0Input = {
+  readonly projectID: Endpoint19_0Request["params"]["projectID"]
+  readonly location?: Endpoint19_0Request["query"]["location"]
+  readonly strategy: Endpoint19_0Request["payload"]["strategy"]
+  readonly directory: Endpoint19_0Request["payload"]["directory"]
+  readonly name?: Endpoint19_0Request["payload"]["name"]
 }
-const Endpoint17_0 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint17_0Input) =>
+const Endpoint19_0 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint19_0Input) =>
   raw["projectCopy.create"]({
     params: { projectID: input["projectID"] },
     query: { location: input["location"] },
     payload: { strategy: input["strategy"], directory: input["directory"], name: input["name"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint17_1Request = Parameters<RawClient["server.projectCopy"]["projectCopy.remove"]>[0]
-type Endpoint17_1Input = {
-  readonly projectID: Endpoint17_1Request["params"]["projectID"]
-  readonly location?: Endpoint17_1Request["query"]["location"]
-  readonly directory: Endpoint17_1Request["payload"]["directory"]
-  readonly force: Endpoint17_1Request["payload"]["force"]
+type Endpoint19_1Request = Parameters<RawClient["server.projectCopy"]["projectCopy.remove"]>[0]
+type Endpoint19_1Input = {
+  readonly projectID: Endpoint19_1Request["params"]["projectID"]
+  readonly location?: Endpoint19_1Request["query"]["location"]
+  readonly directory: Endpoint19_1Request["payload"]["directory"]
+  readonly force: Endpoint19_1Request["payload"]["force"]
 }
-const Endpoint17_1 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint17_1Input) =>
+const Endpoint19_1 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint19_1Input) =>
   raw["projectCopy.remove"]({
     params: { projectID: input["projectID"] },
     query: { location: input["location"] },
     payload: { directory: input["directory"], force: input["force"] },
   }).pipe(Effect.mapError(mapClientError))
 
-type Endpoint17_2Request = Parameters<RawClient["server.projectCopy"]["projectCopy.refresh"]>[0]
-type Endpoint17_2Input = {
-  readonly projectID: Endpoint17_2Request["params"]["projectID"]
-  readonly location?: Endpoint17_2Request["query"]["location"]
+type Endpoint19_2Request = Parameters<RawClient["server.projectCopy"]["projectCopy.refresh"]>[0]
+type Endpoint19_2Input = {
+  readonly projectID: Endpoint19_2Request["params"]["projectID"]
+  readonly location?: Endpoint19_2Request["query"]["location"]
 }
-const Endpoint17_2 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint17_2Input) =>
+const Endpoint19_2 = (raw: RawClient["server.projectCopy"]) => (input: Endpoint19_2Input) =>
   raw["projectCopy.refresh"]({
     params: { projectID: input["projectID"] },
     query: { location: input["location"] },
   }).pipe(Effect.mapError(mapClientError))
 
-const adaptGroup17 = (raw: RawClient["server.projectCopy"]) => ({
-  create: Endpoint17_0(raw),
-  remove: Endpoint17_1(raw),
-  refresh: Endpoint17_2(raw),
+const adaptGroup19 = (raw: RawClient["server.projectCopy"]) => ({
+  create: Endpoint19_0(raw),
+  remove: Endpoint19_1(raw),
+  refresh: Endpoint19_2(raw),
 })
 
 const adaptClient = (raw: RawClient) => ({
@@ -699,18 +776,20 @@ const adaptClient = (raw: RawClient) => ({
   sessions: adaptGroup3(raw["server.session"]),
   messages: adaptGroup4(raw["server.message"]),
   models: adaptGroup5(raw["server.model"]),
-  providers: adaptGroup6(raw["server.provider"]),
-  integrations: adaptGroup7(raw["server.integration"]),
-  credentials: adaptGroup8(raw["server.credential"]),
-  permissions: adaptGroup9(raw["server.permission"]),
-  files: adaptGroup10(raw["server.fs"]),
-  commands: adaptGroup11(raw["server.command"]),
-  skills: adaptGroup12(raw["server.skill"]),
-  events: adaptGroup13(raw["server.event"]),
-  ptys: adaptGroup14(raw["server.pty"]),
-  questions: adaptGroup15(raw["server.question"]),
-  references: adaptGroup16(raw["server.reference"]),
-  projectCopies: adaptGroup17(raw["server.projectCopy"]),
+  generate: adaptGroup6(raw["server.generate"]),
+  providers: adaptGroup7(raw["server.provider"]),
+  integrations: adaptGroup8(raw["server.integration"]),
+  credentials: adaptGroup9(raw["server.credential"]),
+  permissions: adaptGroup10(raw["server.permission"]),
+  files: adaptGroup11(raw["server.fs"]),
+  commands: adaptGroup12(raw["server.command"]),
+  skills: adaptGroup13(raw["server.skill"]),
+  events: adaptGroup14(raw["server.event"]),
+  ptys: adaptGroup15(raw["server.pty"]),
+  "server.shell": adaptGroup16(raw["server.shell"]),
+  questions: adaptGroup17(raw["server.question"]),
+  references: adaptGroup18(raw["server.reference"]),
+  projectCopies: adaptGroup19(raw["server.projectCopy"]),
 })
 
 export const make = (options?: { readonly baseUrl?: URL | string }) =>

packages/client/src/generated/client.ts

@@ -43,6 +43,8 @@ import type {
   MessagesListOutput,
   ModelsListInput,
   ModelsListOutput,
+  GenerateTextInput,
+  GenerateTextOutput,
   ProvidersListInput,
   ProvidersListOutput,
   ProvidersGetInput,
@@ -98,6 +100,16 @@ import type {
   PtysUpdateOutput,
   PtysRemoveInput,
   PtysRemoveOutput,
+  ServerShellListInput,
+  ServerShellListOutput,
+  ServerShellCreateInput,
+  ServerShellCreateOutput,
+  ServerShellGetInput,
+  ServerShellGetOutput,
+  ServerShellOutputInput,
+  ServerShellOutputOutput,
+  ServerShellRemoveInput,
+  ServerShellRemoveOutput,
   QuestionsListRequestsInput,
   QuestionsListRequestsOutput,
   QuestionsListInput,
@@ -399,7 +411,7 @@ export function make(options: ClientOptions) {
             method: "POST",
             path: `/api/session/${encodeURIComponent(input.sessionID)}/compact`,
             successStatus: 204,
-            declaredStatuses: [404, 503, 400, 401],
+            declaredStatuses: [404, 409, 503, 500, 400, 401],
             empty: true,
           },
           requestOptions,
@@ -535,6 +547,21 @@ export function make(options: ClientOptions) {
           requestOptions,
         ),
     },
+    generate: {
+      text: (input: GenerateTextInput, requestOptions?: RequestOptions) =>
+        request<{ readonly data: GenerateTextOutput }>(
+          {
+            method: "POST",
+            path: `/api/generate`,
+            query: { location: input["location"] },
+            body: { prompt: input["prompt"], model: input["model"] },
+            successStatus: 200,
+            declaredStatuses: [400, 503, 401],
+            empty: false,
+          },
+          requestOptions,
+        ).then((value) => value.data),
+    },
     providers: {
       list: (input?: ProvidersListInput, requestOptions?: RequestOptions) =>
         request<ProvidersListOutput>(
@@ -899,6 +926,74 @@ export function make(options: ClientOptions) {
           requestOptions,
         ),
     },
+    "server.shell": {
+      list: (input?: ServerShellListInput, requestOptions?: RequestOptions) =>
+        request<ServerShellListOutput>(
+          {
+            method: "GET",
+            path: `/api/shell`,
+            query: { location: input?.["location"] },
+            successStatus: 200,
+            declaredStatuses: [401, 400],
+            empty: false,
+          },
+          requestOptions,
+        ),
+      create: (input: ServerShellCreateInput, requestOptions?: RequestOptions) =>
+        request<ServerShellCreateOutput>(
+          {
+            method: "POST",
+            path: `/api/shell`,
+            query: { location: input["location"] },
+            body: {
+              command: input["command"],
+              cwd: input["cwd"],
+              timeout: input["timeout"],
+              metadata: input["metadata"],
+            },
+            successStatus: 200,
+            declaredStatuses: [401, 400],
+            empty: false,
+          },
+          requestOptions,
+        ),
+      get: (input: ServerShellGetInput, requestOptions?: RequestOptions) =>
+        request<ServerShellGetOutput>(
+          {
+            method: "GET",
+            path: `/api/shell/${encodeURIComponent(input.id)}`,
+            query: { location: input["location"] },
+            successStatus: 200,
+            declaredStatuses: [404, 401, 400],
+            empty: false,
+          },
+          requestOptions,
+        ),
+      output: (input: ServerShellOutputInput, requestOptions?: RequestOptions) =>
+        request<ServerShellOutputOutput>(
+          {
+            method: "GET",
+            path: `/api/shell/${encodeURIComponent(input.id)}/output`,
+            query: { location: input["location"], cursor: input["cursor"], limit: input["limit"] },
+            successStatus: 200,
+            declaredStatuses: [404, 401, 400],
+            empty: false,
+          },
+          requestOptions,
+        ),
+      remove: (input: ServerShellRemoveInput, requestOptions?: RequestOptions) =>
+        request<ServerShellRemoveOutput>(
+          {
+            method: "DELETE",
+            path: `/api/shell/${encodeURIComponent(input.id)}`,
+            query: { location: input["location"] },
+            successStatus: 204,
+            declaredStatuses: [404, 401, 400],
+            empty: true,
+          },
+          requestOptions,
+        ),
+    },
     questions: {
       listRequests: (input?: QuestionsListRequestsInput, requestOptions?: RequestOptions) =>
         request<QuestionsListRequestsOutput>(

packages/client/src/generated/types.ts

@@ -41,6 +41,14 @@ export type ConflictError = {
 export const isConflictError = (value: unknown): value is ConflictError =>
   typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "ConflictError"
 
+export type SessionBusyError = {
+  readonly _tag: "SessionBusyError"
+  readonly sessionID: string
+  readonly message: string
+}
+export const isSessionBusyError = (value: unknown): value is SessionBusyError =>
+  typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "SessionBusyError"
+
 export type ServiceUnavailableError = {
   readonly _tag: "ServiceUnavailableError"
   readonly message: string
@@ -49,6 +57,14 @@ export type ServiceUnavailableError = {
 export const isServiceUnavailableError = (value: unknown): value is ServiceUnavailableError =>
   typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "ServiceUnavailableError"
 
+export type UnknownError = {
+  readonly _tag: "UnknownError"
+  readonly message: string
+  readonly ref?: string | undefined
+}
+export const isUnknownError = (value: unknown): value is UnknownError =>
+  typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "UnknownError"
+
 export type MessageNotFoundError = {
   readonly _tag: "MessageNotFoundError"
   readonly sessionID: string
@@ -58,22 +74,6 @@ export type MessageNotFoundError = {
 export const isMessageNotFoundError = (value: unknown): value is MessageNotFoundError =>
   typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "MessageNotFoundError"
 
-export type SessionBusyError = {
-  readonly _tag: "SessionBusyError"
-  readonly sessionID: string
-  readonly message: string
-}
-export const isSessionBusyError = (value: unknown): value is SessionBusyError =>
-  typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "SessionBusyError"
-
-export type UnknownError = {
-  readonly _tag: "UnknownError"
-  readonly message: string
-  readonly ref?: string | undefined
-}
-export const isUnknownError = (value: unknown): value is UnknownError =>
-  typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "UnknownError"
-
 export type ProviderNotFoundError = {
   readonly _tag: "ProviderNotFoundError"
   readonly providerID: string
@@ -94,6 +94,10 @@ export type PtyNotFoundError = { readonly _tag: "PtyNotFoundError"; readonly pty
 export const isPtyNotFoundError = (value: unknown): value is PtyNotFoundError =>
   typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "PtyNotFoundError"
 
+export type ShellNotFoundError = { readonly _tag: "ShellNotFoundError"; readonly id: string; readonly message: string }
+export const isShellNotFoundError = (value: unknown): value is ShellNotFoundError =>
+  typeof value === "object" && value !== null && "_tag" in value && value["_tag"] === "ShellNotFoundError"
+
 export type QuestionNotFoundError = {
   readonly _tag: "QuestionNotFoundError"
   readonly requestID: string
@@ -412,7 +416,6 @@ export type SessionsPromptInput = {
         readonly name: string
         readonly source?: { readonly start: number; readonly end: number; readonly text: string }
       }>
-      readonly system?: string
     }
     readonly delivery?: "steer" | "queue" | null
     readonly resume?: boolean | null
@@ -431,7 +434,6 @@ export type SessionsPromptInput = {
         readonly name: string
         readonly source?: { readonly start: number; readonly end: number; readonly text: string }
       }>
-      readonly system?: string
     }
     readonly delivery?: "steer" | "queue" | null
     readonly resume?: boolean | null
@@ -450,7 +452,6 @@ export type SessionsPromptInput = {
         readonly name: string
         readonly source?: { readonly start: number; readonly end: number; readonly text: string }
       }>
-      readonly system?: string
     }
     readonly delivery?: "steer" | "queue" | null
     readonly resume?: boolean | null
@@ -469,7 +470,6 @@ export type SessionsPromptInput = {
         readonly name: string
         readonly source?: { readonly start: number; readonly end: number; readonly text: string }
       }>
-      readonly system?: string
     }
     readonly delivery?: "steer" | "queue" | null
     readonly resume?: boolean | null
@@ -494,7 +494,6 @@ export type SessionsPromptOutput = {
         readonly name: string
         readonly source?: { readonly start: number; readonly end: number; readonly text: string }
       }>
-      readonly system?: string
     }
     readonly delivery: "steer" | "queue"
     readonly timeCreated: number
@@ -574,7 +573,6 @@ export type SessionsContextOutput = {
           readonly name: string
           readonly source?: { readonly start: number; readonly end: number; readonly text: string }
         }>
-        readonly system?: string
         readonly type: "user"
       }
     | {
@@ -775,7 +773,6 @@ export type SessionsHistoryOutput = {
               readonly name: string
               readonly source?: { readonly start: number; readonly end: number; readonly text: string }
             }>
-            readonly system?: string
           }
           readonly delivery: "steer" | "queue"
         }
@@ -803,7 +800,6 @@ export type SessionsHistoryOutput = {
               readonly name: string
               readonly source?: { readonly start: number; readonly end: number; readonly text: string }
             }>
-            readonly system?: string
           }
           readonly delivery: "steer" | "queue"
         }
@@ -1243,7 +1239,6 @@ export type SessionsEventsOutput =
             readonly name: string
             readonly source?: { readonly start: number; readonly end: number; readonly text: string }
           }>
-          readonly system?: string
         }
         readonly delivery: "steer" | "queue"
       }
@@ -1271,7 +1266,6 @@ export type SessionsEventsOutput =
             readonly name: string
             readonly source?: { readonly start: number; readonly end: number; readonly text: string }
           }>
-          readonly system?: string
         }
         readonly delivery: "steer" | "queue"
       }
@@ -1673,7 +1667,6 @@ export type SessionsMessageOutput = {
           readonly name: string
           readonly source?: { readonly start: number; readonly end: number; readonly text: string }
         }>
-        readonly system?: string
         readonly type: "user"
       }
     | {
@@ -1846,7 +1839,6 @@ export type MessagesListOutput = {
           readonly name: string
           readonly source?: { readonly start: number; readonly end: number; readonly text: string }
         }>
-        readonly system?: string
         readonly type: "user"
       }
     | {
@@ -2029,6 +2021,22 @@ export type ModelsListOutput = {
   }>
 }
 
+export type GenerateTextInput = {
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+  }["location"]
+  readonly prompt: {
+    readonly prompt: string
+    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
+  }["prompt"]
+  readonly model?: {
+    readonly prompt: string
+    readonly model?: { readonly id: string; readonly providerID: string; readonly variant?: string } | null
+  }["model"]
+}
+
+export type GenerateTextOutput = { readonly data: { readonly text: string } }["data"]
+
 export type ProvidersListInput = {
   readonly location?: {
     readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
@@ -2730,6 +2738,160 @@ export type PtysRemoveInput = {
 
 export type PtysRemoveOutput = void
 
+export type ServerShellListInput = {
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+  }["location"]
+}
+
+export type ServerShellListOutput = {
+  readonly location: {
+    readonly directory: string
+    readonly workspaceID?: string
+    readonly project: { readonly id: string; readonly directory: string }
+  }
+  readonly data: ReadonlyArray<{
+    readonly id: string
+    readonly status: "running" | "exited" | "timeout" | "killed"
+    readonly command: string
+    readonly cwd: string
+    readonly shell: string
+    readonly file: string
+    readonly pid?: number
+    readonly exit?: number | "Infinity" | "-Infinity" | "NaN"
+    readonly metadata: { readonly [x: string]: JsonValue }
+    readonly time: {
+      readonly started: number | "Infinity" | "-Infinity" | "NaN"
+      readonly completed?: number | "Infinity" | "-Infinity" | "NaN"
+    }
+  }>
+}
+
+export type ServerShellCreateInput = {
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+  }["location"]
+  readonly command: {
+    readonly command: string
+    readonly cwd?: string
+    readonly timeout?: number
+    readonly metadata?: { readonly [x: string]: JsonValue }
+  }["command"]
+  readonly cwd?: {
+    readonly command: string
+    readonly cwd?: string
+    readonly timeout?: number
+    readonly metadata?: { readonly [x: string]: JsonValue }
+  }["cwd"]
+  readonly timeout?: {
+    readonly command: string
+    readonly cwd?: string
+    readonly timeout?: number
+    readonly metadata?: { readonly [x: string]: JsonValue }
+  }["timeout"]
+  readonly metadata?: {
+    readonly command: string
+    readonly cwd?: string
+    readonly timeout?: number
+    readonly metadata?: { readonly [x: string]: JsonValue }
+  }["metadata"]
+}
+
+export type ServerShellCreateOutput = {
+  readonly location: {
+    readonly directory: string
+    readonly workspaceID?: string
+    readonly project: { readonly id: string; readonly directory: string }
+  }
+  readonly data: {
+    readonly id: string
+    readonly status: "running" | "exited" | "timeout" | "killed"
+    readonly command: string
+    readonly cwd: string
+    readonly shell: string
+    readonly file: string
+    readonly pid?: number
+    readonly exit?: number | "Infinity" | "-Infinity" | "NaN"
+    readonly metadata: { readonly [x: string]: JsonValue }
+    readonly time: {
+      readonly started: number | "Infinity" | "-Infinity" | "NaN"
+      readonly completed?: number | "Infinity" | "-Infinity" | "NaN"
+    }
+  }
+}
+
+export type ServerShellGetInput = {
+  readonly id: { readonly id: string }["id"]
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+  }["location"]
+}
+
+export type ServerShellGetOutput = {
+  readonly location: {
+    readonly directory: string
+    readonly workspaceID?: string
+    readonly project: { readonly id: string; readonly directory: string }
+  }
+  readonly data: {
+    readonly id: string
+    readonly status: "running" | "exited" | "timeout" | "killed"
+    readonly command: string
+    readonly cwd: string
+    readonly shell: string
+    readonly file: string
+    readonly pid?: number
+    readonly exit?: number | "Infinity" | "-Infinity" | "NaN"
+    readonly metadata: { readonly [x: string]: JsonValue }
+    readonly time: {
+      readonly started: number | "Infinity" | "-Infinity" | "NaN"
+      readonly completed?: number | "Infinity" | "-Infinity" | "NaN"
+    }
+  }
+}
+
+export type ServerShellOutputInput = {
+  readonly id: { readonly id: string }["id"]
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+    readonly cursor?: number | undefined
+    readonly limit?: number | undefined
+  }["location"]
+  readonly cursor?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+    readonly cursor?: number | undefined
+    readonly limit?: number | undefined
+  }["cursor"]
+  readonly limit?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+    readonly cursor?: number | undefined
+    readonly limit?: number | undefined
+  }["limit"]
+}
+
+export type ServerShellOutputOutput = {
+  readonly location: {
+    readonly directory: string
+    readonly workspaceID?: string
+    readonly project: { readonly id: string; readonly directory: string }
+  }
+  readonly data: {
+    readonly output: string
+    readonly cursor: number
+    readonly size: number
+    readonly truncated: boolean
+  }
+}
+
+export type ServerShellRemoveInput = {
+  readonly id: { readonly id: string }["id"]
+  readonly location?: {
+    readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined
+  }["location"]
+}
+
+export type ServerShellRemoveOutput = void
+
 export type QuestionsListRequestsInput = {
   readonly location?: {
     readonly location?: { readonly directory?: string | undefined; readonly workspace?: string | undefined } | undefined

packages/client/test/promise.test.ts

@@ -11,6 +11,7 @@ test("exposes every standard HTTP API group", () => {
     "sessions",
     "messages",
     "models",
+    "generate",
     "providers",
     "integrations",
     "credentials",

packages/core/src/generate.ts

@@ -0,0 +1,109 @@
+export * as Generate from "./generate"
+
+import { LLM, LLMClient, LLMError } from "@opencode-ai/llm"
+import { Context, Effect, Layer, Schema } from "effect"
+import { Catalog } from "./catalog"
+import { makeLocationNode } from "./effect/app-node"
+import { llmClient } from "./effect/app-node-platform"
+import { Integration } from "./integration"
+import { ModelV2 } from "./model"
+import { SessionRunnerModel } from "./session/runner/model"
+
+export interface TextInput {
+  readonly prompt: string
+  readonly model?: ModelV2.Ref
+}
+
+export class ModelSelectionError extends Schema.TaggedErrorClass<ModelSelectionError>()(
+  "Generate.ModelSelectionError",
+  { message: Schema.String },
+) {}
+
+export class UnavailableError extends Schema.TaggedErrorClass<UnavailableError>()(
+  "Generate.UnavailableError",
+  { message: Schema.String, service: Schema.optional(Schema.String) },
+) {}
+
+export type Error = ModelSelectionError | UnavailableError
+
+export interface Interface {
+  readonly text: (input: TextInput) => Effect.Effect<string, Error>
+}
+
+export class Service extends Context.Service<Service, Interface>()("@opencode/v2/Generate") {}
+
+export const layer = Layer.effect(
+  Service,
+  Effect.gen(function* () {
+    const catalog = yield* Catalog.Service
+    const integrations = yield* Integration.Service
+    const llm = yield* LLMClient.Service
+
+    const selectModel = Effect.fn("Generate.selectModel")(function* (requested?: ModelV2.Ref) {
+      const selected = requested
+        ? yield* catalog.model.get(requested.providerID, requested.id)
+        : yield* catalog.model.default().pipe(
+            Effect.flatMap((model) =>
+              model && SessionRunnerModel.supported(model)
+                ? Effect.succeed(model)
+                : Effect.map(catalog.model.available(), (models) => models.find(SessionRunnerModel.supported)),
+            ),
+          )
+      if (!selected)
+        return yield* new ModelSelectionError({
+          message: requested
+            ? `Model unavailable: ${requested.providerID}/${requested.id}`
+            : "No model specified and no supported model is available",
+        })
+      return yield* SessionRunnerModel.withVariant(selected, requested?.variant).pipe(
+        Effect.mapError(
+          () =>
+            new ModelSelectionError({
+              message: `Variant unavailable for ${selected.providerID}/${selected.id}: ${requested?.variant}`,
+            }),
+        ),
+      )
+    })
+
+    const runText = Effect.fn("Generate.text")(function* (input: TextInput) {
+      const selected = yield* selectModel(input.model)
+      const provider = yield* catalog.provider.get(selected.providerID)
+      const connection = yield* integrations.connection.active(
+        provider?.integrationID ?? Integration.ID.make(selected.providerID),
+      )
+      const credential = connection ? yield* integrations.connection.resolve(connection) : undefined
+      const model = yield* SessionRunnerModel.fromCatalogModel(selected, credential).pipe(
+        Effect.mapError((error) =>
+          input.model
+            ? new ModelSelectionError({ message: error.message })
+            : new UnavailableError({ message: error.message, service: selected.providerID }),
+        ),
+      )
+      const response = yield* llm.generate(LLM.request({ model, prompt: input.prompt })).pipe(
+        Effect.mapError(
+          (error: LLMError) =>
+            new UnavailableError({
+              message: error.message,
+              service: selected.providerID,
+            }),
+        ),
+      )
+      return response.text
+    })
+
+    const text: Interface["text"] = (input) =>
+      runText(input).pipe(
+        Effect.catchTag(
+          "Integration.Authorization",
+          () =>
+            new UnavailableError({
+              message: "Generation credentials are unavailable",
+            }),
+        ),
+      )
+
+    return Service.of({ text })
+  }),
+)
+
+export const node = makeLocationNode({ service: Service, layer, deps: [Catalog.node, Integration.node, llmClient] })

packages/core/src/location-services.ts

@@ -9,6 +9,7 @@ import { Node } from "./effect/app-node"
 import { FileMutation } from "./file-mutation"
 import { FileSystem } from "./filesystem"
 import { FileSystemSearch } from "./filesystem/search"
+import { Generate } from "./generate"
 import { Watcher } from "./filesystem/watcher"
 import { Image } from "./image"
 import { Integration } from "./integration"
@@ -22,11 +23,12 @@ import { Policy } from "./policy"
 import { ProjectCopy } from "./project/copy"
 import { Pty } from "./pty"
 import { QuestionV2 } from "./question"
+import { Shell } from "./shell"
 import { Reference } from "./reference"
 import { ReferenceGuidance } from "./reference/guidance"
 import * as SessionRunnerLLM from "./session/runner/llm"
 import { SessionRunnerModel } from "./session/runner/model"
-import { SessionRuntime } from "./session/runtime"
+import { SessionCompaction } from "./session/compaction"
 import { SessionTodo } from "./session/todo"
 import { SkillV2 } from "./skill"
 import { SkillGuidance } from "./skill/guidance"
@@ -58,6 +60,7 @@ export const locationServices = LayerNode.group([
   FileSystem.node,
   Watcher.node,
   Pty.node,
+  Shell.node,
   SkillV2.node,
   SystemContextRegistry.node,
   SystemContextBuiltIns.node,
@@ -72,11 +75,12 @@ export const locationServices = LayerNode.group([
   ReferenceGuidance.node,
   SessionTodo.node,
   QuestionV2.node,
+  Generate.node,
   ReadToolFileSystem.node,
   BuiltInTools.node,
   SessionRunnerModel.node,
+  SessionCompaction.node,
   Snapshot.node,
-  SessionRuntime.node,
   SessionRunnerLLM.node,
 ])
 

packages/core/src/plugin.ts

@@ -13,10 +13,7 @@ import { Integration } from "./integration"
 import { KeyedMutex } from "./effect/keyed-mutex"
 import { PluginHost } from "./plugin/host"
 import { Reference } from "./reference"
-import { SessionV2 } from "./session"
-import { SessionRuntime } from "./session/runtime"
 import { SkillV2 } from "./skill"
-import { Location } from "./location"
 import { State } from "./state"
 
 export const ID = Plugin.ID
@@ -152,8 +149,6 @@ export const locationLayer = layer.pipe(
   Layer.provideMerge(CommandV2.locationLayer),
   Layer.provideMerge(Integration.locationLayer),
   Layer.provideMerge(Reference.locationLayer),
-  Layer.provideMerge(SessionV2.defaultLayer),
-  Layer.provideMerge(SessionRuntime.layer),
   Layer.provideMerge(SkillV2.locationLayer),
 )
 
@@ -168,9 +163,6 @@ export const node = makeLocationNode({
     CommandV2.node,
     Integration.node,
     Reference.node,
-    SessionV2.node,
-    SessionRuntime.node,
     SkillV2.node,
-    Location.node,
   ],
 })

packages/core/src/plugin/agent.ts

@@ -8,7 +8,9 @@ import { Global } from "../global"
 import { Location } from "../location"
 import { PermissionV2 } from "../permission"
 
-const TRUNCATION_GLOB = path.join(Global.Path.data, "tool-output", "*")
+// Combined output files written by the Shell service, e.g. `<data>/shell/<projectID>/<shellID>.out`.
+// Whitelisted so agents can read a command's full captured output without an external-directory prompt.
+const SHELL_OUTPUT_GLOB = path.join(Global.Path.data, "shell", "*", "*")
 const BUILD_SYSTEM =
   "You are an AI coding agent. Help the user accomplish software engineering tasks by inspecting the workspace, making targeted changes, and using tools according to the configured permissions."
 
@@ -102,7 +104,7 @@ export const Plugin = define({
   effect: Effect.fn(function* (ctx) {
     const location = yield* Location.Service
     const worktree = location.directory
-    const whitelistedDirs = [TRUNCATION_GLOB, path.join(Global.Path.tmp, "*")]
+    const whitelistedDirs = [SHELL_OUTPUT_GLOB, path.join(Global.Path.tmp, "*")]
     const readonlyExternalDirectory: PermissionV2.Ruleset = [
       { action: "external_directory", resource: "*", effect: "ask" },
       ...whitelistedDirs.map(
@@ -124,7 +126,6 @@ export const Plugin = define({
     yield* ctx.agent.transform((draft) => {
       draft.update(AgentV2.defaultID, (item) => {
         item.description = "The default agent. Executes tools based on configured permissions."
-        item.system ??= BUILD_SYSTEM
         item.mode = "primary"
         item.permissions.push(
           ...PermissionV2.merge(defaults, [

packages/core/src/plugin/host.ts

@@ -13,11 +13,7 @@ import { PluginV2 } from "../plugin"
 import { ProviderV2 } from "../provider"
 import { Reference } from "../reference"
 import type { DeepMutable } from "../schema"
-import { SessionV2 } from "../session"
-import { SessionMessage } from "../session/message"
-import { SessionRuntime } from "../session/runtime"
 import { SkillV2 } from "../skill"
-import { Location } from "../location"
 
 const mutable = <T>(value: T) => value as DeepMutable<T>
 
@@ -28,10 +24,7 @@ export const make = Effect.fn("PluginHost.make")(function* (plugin: PluginV2.Int
   const commands = yield* CommandV2.Service
   const integration = yield* Integration.Service
   const reference = yield* Reference.Service
-  const session = yield* SessionV2.Service
-  const sessionRuntime = yield* SessionRuntime.Service
   const skill = yield* SkillV2.Service
-  const location = yield* Location.Service
 
   return {
     options: {},
@@ -212,60 +205,6 @@ export const make = Effect.fn("PluginHost.make")(function* (plugin: PluginV2.Int
           }),
         ),
     },
-    session: {
-      create: (input) =>
-        session.create(
-          input.parentID
-            ? {
-                id: input.id ? SessionV2.ID.make(input.id) : undefined,
-                parentID: SessionV2.ID.make(input.parentID),
-                title: input.title,
-                agent: input.agent ? AgentV2.ID.make(input.agent) : undefined,
-                model: input.model
-                  ? {
-                      id: ModelV2.ID.make(input.model.id),
-                      providerID: ProviderV2.ID.make(input.model.providerID),
-                      variant: input.model.variant ? ModelV2.VariantID.make(input.model.variant) : undefined,
-                    }
-                  : undefined,
-              }
-            : {
-                id: input.id ? SessionV2.ID.make(input.id) : undefined,
-                location: { directory: location.directory, workspaceID: location.workspaceID },
-                title: input.title,
-                agent: input.agent ? AgentV2.ID.make(input.agent) : undefined,
-                model: input.model
-                  ? {
-                      id: ModelV2.ID.make(input.model.id),
-                      providerID: ProviderV2.ID.make(input.model.providerID),
-                      variant: input.model.variant ? ModelV2.VariantID.make(input.model.variant) : undefined,
-                    }
-                  : undefined,
-              },
-        ),
-      get: (sessionID) => session.get(SessionV2.ID.make(sessionID)),
-      messages: (input) =>
-        session.messages({
-          sessionID: SessionV2.ID.make(input.sessionID),
-          limit: input.limit,
-          order: input.order,
-          cursor: input.cursor
-            ? { id: SessionMessage.ID.make(input.cursor.id), direction: input.cursor.direction }
-            : undefined,
-        }),
-      context: (sessionID) => session.context(SessionV2.ID.make(sessionID)),
-      prompt: (input) =>
-        sessionRuntime.prompt({
-          id: input.id ? SessionMessage.ID.make(input.id) : undefined,
-          sessionID: SessionV2.ID.make(input.sessionID),
-          prompt: input.prompt,
-          delivery: input.delivery,
-          resume: input.resume,
-        }),
-      resume: (sessionID) => sessionRuntime.resume(SessionV2.ID.make(sessionID)),
-      wait: (sessionID) => sessionRuntime.wait(SessionV2.ID.make(sessionID)),
-      interrupt: (sessionID) => sessionRuntime.interrupt(SessionV2.ID.make(sessionID)),
-    },
     skill: {
       reload: skill.reload,
       transform: (callback) =>

packages/core/src/plugin/internal.ts

@@ -31,6 +31,7 @@ import { AgentPlugin } from "./agent"
 import { CommandPlugin } from "./command"
 import { ModelsDevPlugin } from "./models-dev"
 import { ProviderPlugins } from "./provider"
+import { SdkPlugins } from "./sdk"
 import { SkillPlugin } from "./skill"
 import { VariantPlugin } from "./variant"
 
@@ -65,6 +66,7 @@ const layer = Layer.effectDiscard(
     const catalog = yield* Catalog.Service
     const commands = yield* CommandV2.Service
     const plugin = yield* PluginV2.Service
+    const sdkPlugins = yield* SdkPlugins.Service
     const integration = yield* Integration.Service
     const agents = yield* AgentV2.Service
     const config = yield* Config.Service
@@ -119,6 +121,8 @@ const layer = Layer.effectDiscard(
         yield* add(ConfigExternalPlugin.Plugin)
         yield* add(ConfigProviderPlugin.Plugin)
         yield* add(VariantPlugin.Plugin)
+        // Embedder-contributed plugins are added last so they layer over config.
+        for (const plugin of sdkPlugins.all()) yield* add(plugin)
       }),
     ).pipe(Effect.withSpan("PluginInternal.boot"), Effect.forkScoped({ startImmediately: true }))
   }),
@@ -151,5 +155,6 @@ export const node = makeLocationNode({
     httpClient,
     SkillV2.node,
     Reference.node,
+    SdkPlugins.node,
   ],
 })

packages/core/src/plugin/promise.ts

@@ -81,16 +81,6 @@ export function fromPromise(plugin: Plugin) {
             transform: transform(host.reference),
             reload: () => run(host.reference.reload()),
           },
-          session: {
-            create: (input) => run(host.session.create(input)),
-            get: (sessionID) => run(host.session.get(sessionID)),
-            messages: (input) => run(host.session.messages(input)),
-            context: (sessionID) => run(host.session.context(sessionID)),
-            prompt: (input) => run(host.session.prompt(input)),
-            resume: (sessionID) => run(host.session.resume(sessionID)),
-            wait: (sessionID) => run(host.session.wait(sessionID)),
-            interrupt: (sessionID) => run(host.session.interrupt(sessionID)),
-          },
           skill: {
             transform: transform(host.skill),
             reload: () => run(host.skill.reload()),

packages/core/src/plugin/sdk.ts

@@ -0,0 +1,37 @@
+export * as SdkPlugins from "./sdk"
+
+import type { Plugin } from "@opencode-ai/plugin/v2/effect"
+import { Context, Effect, Layer } from "effect"
+import { makeGlobalNode } from "../effect/app-node"
+
+/**
+ * Holds the plugins an embedder (the `@opencode-ai/sdk-next` host) contributes,
+ * so `PluginInternal` can add them on every Location boot through the ordinary
+ * `ctx.plugin.add` seam — the same path `ConfigExternalPlugin` uses for plugins
+ * discovered from config. A plugin registered after a Location has booted only
+ * applies to Locations booted afterward, matching config-plugin timing;
+ * embedders register at startup before creating Sessions.
+ *
+ * State lives in this global-node service (like `ApplicationTools`) rather than
+ * module scope, so the list belongs to one embedded instance and is disposed
+ * with it instead of leaking across `OpenCode.create` calls.
+ */
+export interface Interface {
+  readonly register: (plugin: Plugin) => Effect.Effect<void>
+  readonly all: () => readonly Plugin[]
+}
+
+export class Service extends Context.Service<Service, Interface>()("@opencode/SdkPlugins") {}
+
+export const layer = Layer.effect(
+  Service,
+  Effect.sync(() => {
+    const plugins: Plugin[] = []
+    return Service.of({
+      register: (plugin) => Effect.sync(() => void plugins.push(plugin)),
+      all: () => plugins,
+    })
+  }),
+)
+
+export const node = makeGlobalNode({ service: Service, layer, deps: [] })

packages/core/src/pty.ts

@@ -8,7 +8,7 @@ import { Config } from "./config"
 import { EventV2 } from "./event"
 import { Location } from "./location"
 import { PtyID } from "./pty/schema"
-import { Shell } from "./shell"
+import { ShellSelect } from "./shell/select"
 import { lazy } from "./util/lazy"
 
 const BUFFER_LIMIT = 1024 * 1024 * 2
@@ -164,8 +164,8 @@ export const layer = Layer.effect(
 
     const create = Effect.fn("Pty.create")(function* (input: CreateInput) {
       const id = PtyID.ascending()
-      const command = input.command || Shell.preferred(Config.latest(yield* config.entries(), "shell"))
-      const args = Shell.login(command) ? [...(input.args ?? []), "-l"] : [...(input.args ?? [])]
+      const command = input.command || ShellSelect.preferred(Config.latest(yield* config.entries(), "shell"))
+      const args = ShellSelect.login(command) ? [...(input.args ?? []), "-l"] : [...(input.args ?? [])]
       const cwd = input.cwd || location.directory
       const env = {
         ...process.env,

packages/core/src/session.ts

@@ -25,13 +25,15 @@ import { ProjectTable } from "./project/sql"
 import path from "path"
 import { fromRow } from "./session/info"
 import { SessionRunner } from "./session/runner/index"
-import { SessionRuntimeCoordinator } from "./session/runtime-coordinator"
 import { SessionStore } from "./session/store"
+import { SessionExecution } from "./session/execution"
 import { makeGlobalNode } from "./effect/app-node"
+import { LocationServiceMap } from "./location-service-map"
 import { MessageDecodeError } from "./session/error"
 import { SessionEvent } from "./session/event"
 import { SessionInput } from "./session/input"
 import { Snapshot } from "./snapshot"
+import { SessionCompaction } from "./session/compaction"
 import { SessionRevert } from "./session/revert"
 import { Revert } from "@opencode-ai/schema/revert"
 import { FSUtil } from "./fs-util"
@@ -86,7 +88,6 @@ type CreateInput = CreateBaseInput &
 
 type CompactInput = {
   sessionID: SessionSchema.ID
-  prompt?: Prompt
 }
 
 export class NotFoundError extends Schema.TaggedErrorClass<NotFoundError>()("Session.NotFoundError", {
@@ -112,7 +113,7 @@ export class BusyError extends Schema.TaggedErrorClass<BusyError>()("Session.Bus
 export const MessageNotFoundError = SessionRevert.MessageNotFoundError
 export type MessageNotFoundError = SessionRevert.MessageNotFoundError
 
-export type Error = NotFoundError | MessageDecodeError | OperationUnavailableError | PromptConflictError
+export type Error = NotFoundError | MessageDecodeError | OperationUnavailableError | PromptConflictError | BusyError
 
 export interface Interface {
   readonly list: (input?: ListInput) => Effect.Effect<SessionSchema.Info[]>
@@ -168,7 +169,9 @@ export interface Interface {
     skill: string
     resume?: boolean
   }) => Effect.Effect<void, OperationUnavailableError>
-  readonly compact: (input: CompactInput) => Effect.Effect<void, NotFoundError | OperationUnavailableError>
+  readonly compact: (
+    input: CompactInput,
+  ) => Effect.Effect<void, NotFoundError | BusyError | MessageDecodeError | OperationUnavailableError>
   readonly wait: (id: SessionSchema.ID) => Effect.Effect<void, NotFoundError>
   readonly active: Effect.Effect<ReadonlySet<SessionSchema.ID>>
   readonly resume: (sessionID: SessionSchema.ID) => Effect.Effect<void, NotFoundError | SessionRunner.RunError>
@@ -193,8 +196,9 @@ export const layer = Layer.effect(
     const db = database.db
     const events = yield* EventV2.Service
     const projects = yield* ProjectV2.Service
-    const runtimeCoordinator = yield* SessionRuntimeCoordinator.Service
+    const execution = yield* SessionExecution.Service
     const store = yield* SessionStore.Service
+    const locations = yield* LocationServiceMap.Service
     const decodeMessage = Schema.decodeUnknownEffect(SessionMessage.Message)
     const isDurableSessionEvent = Schema.is(SessionEvent.Durable)
     const decode = (row: typeof SessionMessageTable.$inferSelect) =>
@@ -368,7 +372,11 @@ export const layer = Layer.effect(
       prompt: Effect.fn("V2Session.prompt")((input) =>
         Effect.uninterruptible(
           Effect.gen(function* () {
-            yield* result.get(input.sessionID)
+            const session = yield* result.get(input.sessionID)
+            // A staged revert must be committed before admitting new input so the prompt
+            // continues from the reverted boundary rather than stale post-boundary history.
+            if (session.revert)
+              yield* SessionRevert.commit(session).pipe(Effect.provideService(EventV2.Service, events))
             const prompt = resolvePrompt(input.prompt)
             const messageID = input.id ?? SessionMessage.ID.create()
             const delivery = input.delivery ?? "steer"
@@ -387,8 +395,7 @@ export const layer = Layer.effect(
             )
             if (!SessionInput.equivalent(admitted, expected))
               return yield* new PromptConflictError({ sessionID: input.sessionID, messageID })
-            if (input.resume !== false)
-              return yield* Effect.die("SessionV2.prompt with resume moved to SessionRuntime.Service")
+            if (input.resume !== false) yield* execution.wake(admitted.sessionID)
             return admitted
           }),
         ),
@@ -432,30 +439,54 @@ export const layer = Layer.effect(
         })
       }),
       compact: Effect.fn("V2Session.compact")(function* (input) {
-        yield* result.get(input.sessionID)
-        return yield* new OperationUnavailableError({ operation: "compact" })
+        const session = yield* result.get(input.sessionID)
+        // TODO: admit manual compaction as durable pending work, like prompt input, instead of rejecting active sessions.
+        if ((yield* execution.active).has(input.sessionID)) return yield* new BusyError({ sessionID: input.sessionID })
+        const context = yield* store.context(input.sessionID)
+        const compacted = yield* Effect.gen(function* () {
+          const compaction = yield* SessionCompaction.Service
+          return yield* compaction.compactManual({ session, messages: context })
+        }).pipe(
+          Effect.provide(locations.get(session.location)),
+          Effect.catch(() => Effect.succeed(false)),
+        )
+        if (!compacted) return yield* new OperationUnavailableError({ operation: "compact" })
+        return undefined
       }),
       wait: Effect.fn("V2Session.wait")(function* (sessionID) {
         yield* result.get(sessionID)
-        return yield* Effect.die("SessionV2.wait moved to SessionRuntime.Service")
+        yield* execution.awaitIdle(sessionID)
       }),
-      active: runtimeCoordinator.active,
+      active: execution.active,
       resume: Effect.fn("V2Session.resume")(function* (sessionID) {
         yield* result.get(sessionID)
-        return yield* Effect.die("SessionV2.resume moved to SessionRuntime.Service")
+        yield* execution.resume(sessionID)
       }),
-      interrupt: Effect.fn("V2Session.interrupt")(() => Effect.die("SessionV2.interrupt moved to SessionRuntime.Service")),
+      interrupt: Effect.fn("V2Session.interrupt")((sessionID) =>
+        Effect.uninterruptible(execution.interrupt(sessionID)),
+      ),
       revert: {
         stage: Effect.fn("V2Session.revert.stage")(function* (input) {
-          yield* result.get(input.sessionID)
-          return yield* Effect.die("SessionV2.revert.stage moved to SessionRuntime.Service")
+          const session = yield* result.get(input.sessionID)
+          if ((yield* execution.active).has(input.sessionID))
+            return yield* new BusyError({ sessionID: input.sessionID })
+          return yield* SessionRevert.stage({ session, messageID: input.messageID, files: input.files }).pipe(
+            Effect.provideService(Database.Service, database),
+            Effect.provideService(EventV2.Service, events),
+            Effect.provide(locations.get(session.location)),
+          )
         }),
         clear: Effect.fn("V2Session.revert.clear")(function* (sessionID) {
-          yield* result.get(sessionID)
-          return yield* Effect.die("SessionV2.revert.clear moved to SessionRuntime.Service")
+          const session = yield* result.get(sessionID)
+          if ((yield* execution.active).has(sessionID)) return yield* new BusyError({ sessionID })
+          return yield* SessionRevert.clear(session).pipe(
+            Effect.provideService(EventV2.Service, events),
+            Effect.provide(locations.get(session.location)),
+          )
         }),
         commit: Effect.fn("V2Session.revert.commit")(function* (sessionID) {
           const session = yield* result.get(sessionID)
+          if ((yield* execution.active).has(sessionID)) return yield* new BusyError({ sessionID })
           return yield* SessionRevert.commit(session).pipe(Effect.provideService(EventV2.Service, events))
         }),
       },
@@ -471,7 +502,6 @@ export const defaultLayer = layer.pipe(
   Layer.provide(EventV2.defaultLayer),
   Layer.provide(Database.defaultLayer),
   Layer.provide(ProjectV2.defaultLayer),
-  Layer.provide(SessionRuntimeCoordinator.defaultLayer),
   Layer.orDie,
 )
 
@@ -479,7 +509,6 @@ const resolvePrompt = (input: PromptInput.Prompt) =>
   Prompt.make({
     text: input.text,
     agents: input.agents,
-    system: input.system,
     files: input.files?.map((file) => {
       const dataMime = file.uri.match(/^data:([^;,]+)[;,]/i)?.[1]
       const target = URL.canParse(file.uri) ? new URL(file.uri).pathname : (file.name ?? file.uri)
@@ -497,8 +526,9 @@ export const node = makeGlobalNode({
     Database.node,
     EventV2.node,
     ProjectV2.node,
-    SessionRuntimeCoordinator.node,
+    SessionExecution.node,
     SessionStore.node,
+    LocationServiceMap.node,
     SessionProjector.node,
   ],
 })

packages/core/src/session/compaction.ts

@@ -1,11 +1,16 @@
 export * as SessionCompaction from "./compaction"
 
-import { LLM, LLMError, LLMEvent, Message, type LLMRequest, type Model } from "@opencode-ai/llm"
-import { DateTime, Effect, Stream } from "effect"
+import { LLM, LLMClient, LLMError, LLMEvent, Message, type LLMRequest, type Model } from "@opencode-ai/llm"
+import { Context, DateTime, Effect, Layer, Stream } from "effect"
 import type { Config } from "../config"
+import { Config as ConfigV2 } from "../config"
 import type { EventV2 } from "../event"
+import { EventV2 as EventV2Service } from "../event"
+import { makeLocationNode } from "../effect/app-node"
+import { llmClient } from "../effect/app-node-platform"
 import { SessionEvent } from "./event"
 import { SessionMessage } from "./message"
+import { SessionRunnerModel } from "./runner/model"
 import { SessionSchema } from "./schema"
 import { Token } from "../util/token"
 
@@ -50,11 +55,6 @@ Rules:
 - Preserve exact file paths, commands, error strings, and identifiers when known.
 - Do not mention the summary process or that context was compacted.`
 
-type Entry = {
-  readonly seq: number
-  readonly message: SessionMessage.Message
-}
-
 type Settings = {
   readonly auto: boolean
   readonly buffer: number
@@ -69,13 +69,31 @@ type Dependencies = {
   readonly config: readonly Config.Entry[]
 }
 
-type Input = {
+export type AutoInput = {
   readonly sessionID: SessionSchema.ID
-  readonly entries: readonly Entry[]
-  readonly model: Model
+  readonly messages: readonly SessionMessage.Message[]
   readonly request: LLMRequest
 }
 
+type CompactInput = {
+  readonly sessionID: SessionSchema.ID
+  readonly messages: readonly SessionMessage.Message[]
+  readonly model: Model
+}
+
+export type ManualInput = {
+  readonly session: SessionSchema.Info
+  readonly messages: readonly SessionMessage.Message[]
+}
+
+export interface Interface {
+  readonly compactIfNeeded: (input: AutoInput) => Effect.Effect<boolean>
+  readonly compactAfterOverflow: (input: AutoInput) => Effect.Effect<boolean>
+  readonly compactManual: (input: ManualInput) => Effect.Effect<boolean>
+}
+
+export class Service extends Context.Service<Service, Interface>()("@opencode/v2/SessionCompaction") {}
+
 const estimate = (value: unknown) => Token.estimate(JSON.stringify(value))
 
 const truncate = (value: string) =>
@@ -131,14 +149,14 @@ const settings = (documents: readonly Config.Entry[]) => {
 }
 
 const select = (
-  entries: readonly Entry[],
+  messages: readonly SessionMessage.Message[],
   tokens: number,
 ): { readonly head: string; readonly recent: string } | undefined => {
-  const conversation = entries
-    .filter((entry) => entry.message.type !== "compaction")
-    .map((entry) => serialize(entry.message))
+  const conversation = messages
+    .filter((message) => message.type !== "compaction")
+    .map(serialize)
     .filter(Boolean)
-  if (conversation.length === 0) return
+  if (conversation.length === 0) return undefined
   let total = 0
   let split = conversation.length
   let splitPrefix = ""
@@ -172,19 +190,21 @@ export const buildPrompt = (input: { readonly previousSummary?: string; readonly
     ...input.context,
   ].join("\n\n")
 
-export const make = (dependencies: Dependencies) => {
+const make = (dependencies: Dependencies) => {
   const config = settings(dependencies.config)
-  const compactAfterOverflow = Effect.fn("SessionCompaction.compactAfterOverflow")(function* (input: Input) {
+  const compact = Effect.fn("SessionCompaction.compact")(function* (input: {
+    readonly sessionID: SessionSchema.ID
+    readonly model: Model
+    readonly reason: SessionMessage.Compaction["reason"]
+    readonly previousSummary?: string
+    readonly context: readonly string[]
+    readonly recent: string
+    readonly output?: number
+  }) {
     const context = input.model.route.defaults.limits?.context
     if (context === undefined || context <= 0) return false
-    const output = input.request.generation?.maxTokens ?? input.model.route.defaults.limits?.output ?? 0
-    const selected = select(input.entries, config.tokens)
-    const previousSummary = input.entries.find((entry) => entry.message.type === "compaction")?.message
-    if (!selected || (selected.head.length === 0 && previousSummary?.type !== "compaction")) return false
-    const summaryPrompt = buildPrompt({
-      previousSummary: previousSummary?.type === "compaction" ? previousSummary.summary : undefined,
-      context: [previousSummary?.type === "compaction" ? previousSummary.recent : "", selected.head].filter(Boolean),
-    })
+    const output = input.output ?? input.model.route.defaults.limits?.output ?? 0
+    const summaryPrompt = buildPrompt({ previousSummary: input.previousSummary, context: input.context })
     const summaryOutput = Math.min(output || SUMMARY_OUTPUT_TOKENS, SUMMARY_OUTPUT_TOKENS)
     if (Token.estimate(summaryPrompt) > context - summaryOutput) return false
     const messageID = SessionMessage.ID.create()
@@ -192,7 +212,7 @@ export const make = (dependencies: Dependencies) => {
       sessionID: input.sessionID,
       messageID,
       timestamp: yield* DateTime.now,
-      reason: "auto",
+      reason: input.reason,
     })
 
     const chunks: string[] = []
@@ -221,17 +241,58 @@ export const make = (dependencies: Dependencies) => {
       sessionID: input.sessionID,
       messageID,
       timestamp: yield* DateTime.now,
-      reason: "auto",
+      reason: input.reason,
       text: summary,
-      recent: selected.recent,
+      recent: input.recent,
     })
     return true
   })
-  const compactIfNeeded = Effect.fn("SessionCompaction.compactIfNeeded")(function* (input: Input) {
-    if (!config.auto) return false
+  const compactAfterOverflow = Effect.fn("SessionCompaction.compactAfterOverflow")(function* (input: AutoInput) {
+    return yield* compactSelected({
+      sessionID: input.sessionID,
+      messages: input.messages,
+      model: input.request.model,
+      reason: "auto",
+      force: false,
+      output: input.request.generation?.maxTokens ?? input.request.model.route.defaults.limits?.output ?? 0,
+    })
+  })
+  const compactSelected = Effect.fn("SessionCompaction.compactSelected")(function* (
+    input: CompactInput & {
+      readonly reason: SessionMessage.Compaction["reason"]
+      readonly force: boolean
+      readonly output?: number
+    },
+  ) {
     const context = input.model.route.defaults.limits?.context
     if (context === undefined || context <= 0) return false
-    const output = input.request.generation?.maxTokens ?? input.model.route.defaults.limits?.output ?? 0
+    const selected = select(input.messages, config.tokens)
+    if (!selected) return false
+    const previousSummary = input.messages.find((message) => message.type === "compaction")
+    const hasHead = selected.head.length > 0
+    if (!hasHead && previousSummary?.type !== "compaction" && !input.force) return false
+    const forcedShortContext = input.force && !hasHead
+    const previousRecent = previousSummary?.type === "compaction" ? previousSummary.recent : ""
+    return yield* compact({
+      sessionID: input.sessionID,
+      model: input.model,
+      reason: input.reason,
+      previousSummary: previousSummary?.type === "compaction" ? previousSummary.summary : undefined,
+      context: (forcedShortContext ? [previousRecent, selected.recent] : [previousRecent, selected.head]).filter(
+        Boolean,
+      ),
+      recent: forcedShortContext ? "" : selected.recent,
+      output: input.output,
+    })
+  })
+  const compactManual = Effect.fn("SessionCompaction.compactManual")(function* (input: CompactInput) {
+    return yield* compactSelected({ ...input, reason: "manual", force: true })
+  })
+  const compactIfNeeded = Effect.fn("SessionCompaction.compactIfNeeded")(function* (input: AutoInput) {
+    if (!config.auto) return false
+    const context = input.request.model.route.defaults.limits?.context
+    if (context === undefined || context <= 0) return false
+    const output = input.request.generation?.maxTokens ?? input.request.model.route.defaults.limits?.output ?? 0
     if (
       estimate({ system: input.request.system, messages: input.request.messages, tools: input.request.tools }) <=
       context - Math.max(output, config.buffer)
@@ -242,5 +303,37 @@ export const make = (dependencies: Dependencies) => {
   return {
     compactIfNeeded,
     compactAfterOverflow,
+    compactManual,
   }
 }
+
+export const layer = Layer.effect(
+  Service,
+  Effect.gen(function* () {
+    const events = yield* EventV2Service.Service
+    const llm = yield* LLMClient.Service
+    const config = yield* ConfigV2.Service
+    const models = yield* SessionRunnerModel.Service
+    const compaction = make({ events, llm, config: yield* config.entries() })
+
+    return Service.of({
+      compactIfNeeded: compaction.compactIfNeeded,
+      compactAfterOverflow: compaction.compactAfterOverflow,
+      compactManual: Effect.fn("SessionCompaction.compactManual")(function* (input) {
+        const model = yield* models.resolve(input.session).pipe(Effect.catch(() => Effect.succeed(undefined)))
+        if (!model) return false
+        return yield* compaction.compactManual({
+          sessionID: input.session.id,
+          messages: input.messages,
+          model,
+        })
+      }),
+    })
+  }),
+)
+
+export const node = makeLocationNode({
+  service: Service,
+  layer,
+  deps: [EventV2Service.node, llmClient, ConfigV2.node, SessionRunnerModel.node],
+})

packages/core/src/session/message-updater.ts

@@ -133,7 +133,6 @@ export function update(adapter: Adapter, event: SessionEvent.Event) {
             text: event.data.prompt.text,
             files: event.data.prompt.files,
             agents: event.data.prompt.agents,
-            system: event.data.prompt.system,
             time: { created: event.data.timestamp },
           }),
         )

packages/core/src/session/runner/llm.ts

@@ -10,7 +10,6 @@ import {
 } from "@opencode-ai/llm"
 import { Cause, DateTime, Effect, FiberSet, Layer, Option, Semaphore, Stream } from "effect"
 import { AgentV2 } from "../../agent"
-import { Config } from "../../config"
 import { Database } from "../../database/database"
 import { EventV2 } from "../../event"
 import { Location } from "../../location"
@@ -35,6 +34,7 @@ import { SessionRunnerModel } from "./model"
 import { createLLMEventPublisher } from "./publish-llm-event"
 import { toLLMMessages } from "./to-llm-message"
 import { MAX_STEPS_PROMPT } from "./max-steps"
+import { SessionRunnerSystemPrompt } from "./system-prompt"
 import { Snapshot } from "../../snapshot"
 import { makeLocationNode } from "../../effect/app-node"
 import { llmClient } from "../../effect/app-node-platform"
@@ -102,10 +102,9 @@ export const layer = Layer.effect(
     const systemContext = yield* SystemContextRegistry.Service
     const skillGuidance = yield* SkillGuidance.Service
     const referenceGuidance = yield* ReferenceGuidance.Service
-    const config = yield* Config.Service
     const snapshots = yield* Snapshot.Service
     const db = (yield* Database.Service).db
-    const compaction = SessionCompaction.make({ events, llm, config: yield* config.entries() })
+    const compaction = yield* SessionCompaction.Service
     const getSession = Effect.fn("SessionRunner.getSession")(function* (sessionID: SessionSchema.ID) {
       const session = yield* store.get(sessionID)
       if (!session) return yield* Effect.die(`Session not found: ${sessionID}`)
@@ -194,23 +193,20 @@ export const layer = Layer.effect(
       const model = yield* models.resolve(session)
       const entries = yield* SessionHistory.entriesForRunner(db, session.id, system.baselineSeq)
       const context = entries.map((entry) => entry.message)
-      // Mirror V1 (session/llm/request.ts): append the current turn's per-request system string after the
-      // agent prompt and durable baseline. The current turn's user prompt is the latest user message in context.
-      const turnSystem = context.findLast((message) => message.type === "user")?.system
       const isLastStep = agent.info?.steps !== undefined && currentStep >= agent.info.steps
       const toolMaterialization = isLastStep ? undefined : yield* tools.materialize(agent.info?.permissions)
       const promptCacheKey = /^ses_[0-9a-f]{64}$/.test(session.id) ? session.id.slice(4) : session.id
       const request = LLM.request({
         model,
         providerOptions: { openai: { promptCacheKey } },
-        system: [agent.info?.system, system.baseline, turnSystem]
+        system: [agent.info?.system ? agent.info.system : SessionRunnerSystemPrompt.provider(model), system.baseline]
           .filter((part): part is string => part !== undefined && part.length > 0)
           .map(SystemPart.make),
         messages: [...toLLMMessages(context, model), ...(isLastStep ? [Message.assistant(MAX_STEPS_PROMPT)] : [])],
         tools: toolMaterialization?.definitions ?? [],
         toolChoice: isLastStep ? "none" : undefined,
       })
-      if (yield* compaction.compactIfNeeded({ sessionID: session.id, entries, model, request }))
+      if (yield* compaction.compactIfNeeded({ sessionID: session.id, messages: context, request }))
         return yield* Effect.die(continueAfterCompaction(currentStep))
       const startSnapshot = yield* snapshots.capture()
       const publisher = createLLMEventPublisher(events, {
@@ -223,9 +219,9 @@ export const layer = Layer.effect(
         },
         snapshot: startSnapshot,
       })
-      const withPublication = Semaphore.makeUnsafe(1).withPermit
+      const publication = Semaphore.makeUnsafe(1)
       const publish = (event: LLMEvent, outputPaths: ReadonlyArray<string> = []) =>
-        withPublication(publisher.publish(event, outputPaths))
+        publication.withPermit(publisher.publish(event, outputPaths))
       let overflowFailure: ProviderErrorEvent | undefined
       const providerStream = llm.stream(request).pipe(
         Stream.runForEach((event) =>
@@ -240,7 +236,9 @@ export const layer = Layer.effect(
             yield* publish(event)
             if (event.type !== "tool-call" || event.providerExecuted) return
             if (!toolMaterialization) {
-              yield* withPublication(publisher.failUnsettledTools("Tools are disabled after the maximum agent steps"))
+              yield* publication.withPermit(
+                publisher.failUnsettledTools("Tools are disabled after the maximum agent steps"),
+              )
               return
             }
             needsContinuation = true
@@ -269,7 +267,7 @@ export const layer = Layer.effect(
             ).pipe(FiberSet.run(toolFibers))
           }),
         ),
-        Effect.ensuring(withPublication(publisher.flush())),
+        Effect.ensuring(publication.withPermit(publisher.flush())),
       )
 
       return yield* Effect.uninterruptibleMask((restore) =>
@@ -281,14 +279,14 @@ export const layer = Layer.effect(
             recoverOverflow &&
             !publisher.hasAssistantStarted() &&
             isContextOverflowFailure(overflowFailure ?? failure) &&
-            (yield* restore(recoverOverflow({ sessionID: session.id, entries, model, request })))
+            (yield* restore(recoverOverflow({ sessionID: session.id, messages: context, request })))
           )
             return yield* Effect.die(continueAfterOverflowCompaction(currentStep))
           if (overflowFailure) yield* publish(overflowFailure)
           const llmFailure = failure instanceof LLMError ? failure : undefined
           if (llmFailure && !publisher.hasProviderError()) {
-            yield* withPublication(publisher.failUnsettledTools("Provider did not return a tool result", true))
-            yield* withPublication(publisher.failAssistant(llmFailure.reason.message))
+            yield* publication.withPermit(publisher.failUnsettledTools("Provider did not return a tool result", true))
+            yield* publication.withPermit(publisher.failAssistant(llmFailure.reason.message))
           }
           if (stream._tag === "Failure" && Cause.hasInterrupts(stream.cause)) yield* FiberSet.clear(toolFibers)
           const settled = yield* restore(awaitToolFibers(toolFibers)).pipe(Effect.exit)
@@ -296,19 +294,19 @@ export const layer = Layer.effect(
           const toolsInterrupted = settled._tag === "Failure" && Cause.hasInterrupts(settled.cause)
           if (settled._tag === "Failure" && isQuestionRejected(settled.cause)) {
             yield* FiberSet.clear(toolFibers)
-            yield* withPublication(publisher.failUnsettledTools("Tool execution interrupted"))
-            yield* withPublication(publisher.failAssistant("Provider turn interrupted"))
+            yield* publication.withPermit(publisher.failUnsettledTools("Tool execution interrupted"))
+            yield* publication.withPermit(publisher.failAssistant("Provider turn interrupted"))
             return yield* Effect.interrupt
           }
           if (streamInterrupted || toolsInterrupted) {
             yield* FiberSet.clear(toolFibers)
-            yield* withPublication(publisher.failUnsettledTools("Tool execution interrupted"))
-            yield* withPublication(publisher.failAssistant("Provider turn interrupted"))
+            yield* publication.withPermit(publisher.failUnsettledTools("Tool execution interrupted"))
+            yield* publication.withPermit(publisher.failAssistant("Provider turn interrupted"))
           }
           if (settled._tag === "Failure" && !Cause.hasInterrupts(settled.cause)) {
             const failure = Cause.squash(settled.cause)
             const message = failure instanceof Error ? failure.message : String(failure)
-            yield* withPublication(publisher.failUnsettledTools(`Tool execution failed: ${message}`))
+            yield* publication.withPermit(publisher.failUnsettledTools(`Tool execution failed: ${message}`))
           }
           const stepSettlement = publisher.stepSettlement()
           if (stepSettlement && !streamInterrupted && !toolsInterrupted && !publisher.hasProviderError()) {
@@ -319,7 +317,7 @@ export const layer = Layer.effect(
                     .files({ from: startSnapshot, to: endSnapshot })
                     .pipe(Effect.catch(() => Effect.succeed(undefined)))
                 : undefined
-            yield* withPublication(
+            yield* publication.withPermit(
               events.publish(SessionEvent.Step.Ended, {
                 sessionID: session.id,
                 timestamp: yield* DateTime.now,
@@ -333,9 +331,9 @@ export const layer = Layer.effect(
             )
           }
           if (publisher.hasProviderError())
-            yield* withPublication(publisher.failUnsettledTools("Tool execution interrupted"))
+            yield* publication.withPermit(publisher.failUnsettledTools("Tool execution interrupted"))
           if (stream._tag === "Success" && !publisher.hasProviderError())
-            yield* withPublication(publisher.failUnsettledTools("Provider did not return a tool result", true))
+            yield* publication.withPermit(publisher.failUnsettledTools("Provider did not return a tool result", true))
           if (stream._tag === "Failure") return yield* Effect.failCause(stream.cause)
           if (settled._tag === "Failure" && Cause.hasInterrupts(settled.cause))
             return yield* Effect.failCause(settled.cause)
@@ -424,7 +422,7 @@ export const node = makeLocationNode({
     SystemContextRegistry.node,
     SkillGuidance.node,
     ReferenceGuidance.node,
-    Config.node,
+    SessionCompaction.node,
     Snapshot.node,
     Database.node,
   ],

packages/core/src/session/runner/model.ts

@@ -101,7 +101,7 @@ const withDefaults = (model: ModelV2.Info, route: AnyRoute) => {
   })
 }
 
-const withVariant = (
+export const withVariant = (
   model: ModelV2.Info,
   variantID: ModelV2.VariantID | undefined,
 ): Effect.Effect<ModelV2.Info, VariantUnavailableError> => {

packages/core/src/session/runner/prompt/anthropic.txt

@@ -0,0 +1,105 @@
+You are OpenCode, the best coding agent on the planet.
+
+You are an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
+
+IMPORTANT: You must NEVER generate or guess URLs for the user unless you are confident that the URLs are for helping the user with programming. You may use URLs provided by the user in their messages or local files.
+
+If the user asks for help or wants to give feedback inform them of the following:
+- ctrl+p to list available actions
+- To give feedback, users should report the issue at
+  https://github.com/anomalyco/opencode
+
+When the user directly asks about OpenCode (eg. "can OpenCode do...", "does OpenCode have..."), or asks in second person (eg. "are you able...", "can you do..."), or asks how to use a specific OpenCode feature (eg. implement a hook, write a slash command, or install an MCP server), use the WebFetch tool to gather information to answer the question from OpenCode docs. The list of available docs is available at https://opencode.ai/docs
+
+# Tone and style
+- Only use emojis if the user explicitly requests it. Avoid using emojis in all communication unless asked.
+- Your output will be displayed on a command line interface. Your responses should be short and concise. You can use GitHub-flavored markdown for formatting, and will be rendered in a monospace font using the CommonMark specification.
+- Output text to communicate with the user; all text you output outside of tool use is displayed to the user. Only use tools to complete tasks. Never use tools like Bash or code comments as means to communicate with the user during the session.
+- NEVER create files unless they're absolutely necessary for achieving your goal. ALWAYS prefer editing an existing file to creating a new one. This includes markdown files.
+
+# Professional objectivity
+Prioritize technical accuracy and truthfulness over validating the user's beliefs. Focus on facts and problem-solving, providing direct, objective technical info without any unnecessary superlatives, praise, or emotional validation. It is best for the user if OpenCode honestly applies the same rigorous standards to all ideas and disagrees when necessary, even if it may not be what the user wants to hear. Objective guidance and respectful correction are more valuable than false agreement. Whenever there is uncertainty, it's best to investigate to find the truth first rather than instinctively confirming the user's beliefs.
+
+# Task Management
+You have access to the TodoWrite tools to help you manage and plan tasks. Use these tools VERY frequently to ensure that you are tracking your tasks and giving the user visibility into your progress.
+These tools are also EXTREMELY helpful for planning tasks, and for breaking down larger complex tasks into smaller steps. If you do not use this tool when planning, you may forget to do important tasks - and that is unacceptable.
+
+It is critical that you mark todos as completed as soon as you are done with a task. Do not batch up multiple tasks before marking them as completed.
+
+Examples:
+
+<example>
+user: Run the build and fix any type errors
+assistant: I'm going to use the TodoWrite tool to write the following items to the todo list:
+- Run the build
+- Fix any type errors
+
+I'm now going to run the build using Bash.
+
+Looks like I found 10 type errors. I'm going to use the TodoWrite tool to write 10 items to the todo list.
+
+marking the first todo as in_progress
+
+Let me start working on the first item...
+
+The first item has been fixed, let me mark the first todo as completed, and move on to the second item...
+..
+..
+</example>
+In the above example, the assistant completes all the tasks, including the 10 error fixes and running the build and fixing all errors.
+
+<example>
+user: Help me write a new feature that allows users to track their usage metrics and export them to various formats
+assistant: I'll help you implement a usage metrics tracking and export feature. Let me first use the TodoWrite tool to plan this task.
+Adding the following todos to the todo list:
+1. Research existing metrics tracking in the codebase
+2. Design the metrics collection system
+3. Implement core metrics tracking functionality
+4. Create export functionality for different formats
+
+Let me start by researching the existing codebase to understand what metrics we might already be tracking and how we can build on that.
+
+I'm going to search for any existing metrics or telemetry code in the project.
+
+I've found some existing telemetry code. Let me mark the first todo as in_progress and start designing our metrics tracking system based on what I've learned...
+
+[Assistant continues implementing the feature step by step, marking todos as in_progress and completed as they go]
+</example>
+
+
+# Doing tasks
+The user will primarily request you perform software engineering tasks. This includes solving bugs, adding new functionality, refactoring code, explaining code, and more. For these tasks the following steps are recommended:
+- 
+- Use the TodoWrite tool to plan the task if required
+
+- Tool results and user messages may include <system-reminder> tags. <system-reminder> tags contain useful information and reminders. They are automatically added by the system, and bear no direct relation to the specific tool results or user messages in which they appear.
+
+
+# Tool usage policy
+- When doing file search, prefer to use the Task tool in order to reduce context usage.
+- You should proactively use the Task tool with specialized agents when the task at hand matches the agent's description.
+
+- When WebFetch returns a message about a redirect to a different host, you should immediately make a new WebFetch request with the redirect URL provided in the response.
+- You can call multiple tools in a single response. If you intend to call multiple tools and there are no dependencies between them, make all independent tool calls in parallel. Maximize use of parallel tool calls where possible to increase efficiency. However, if some tool calls depend on previous calls to inform dependent values, do NOT call these tools in parallel and instead call them sequentially. For instance, if one operation must complete before another starts, run these operations sequentially instead. Never use placeholders or guess missing parameters in tool calls.
+- If the user specifies that they want you to run tools "in parallel", you MUST send a single message with multiple tool use content blocks. For example, if you need to launch multiple agents in parallel, send a single message with multiple Task tool calls.
+- Use specialized tools instead of bash commands when possible, as this provides a better user experience. For file operations, use dedicated tools: Read for reading files instead of cat/head/tail, Edit for editing instead of sed/awk, and Write for creating files instead of cat with heredoc or echo redirection. Reserve bash tools exclusively for actual system commands and terminal operations that require shell execution. NEVER use bash echo or other command-line tools to communicate thoughts, explanations, or instructions to the user. Output all communication directly in your response text instead.
+- VERY IMPORTANT: When exploring the codebase to gather context or to answer a question that is not a needle query for a specific file/class/function, it is CRITICAL that you use the Task tool instead of running search commands directly.
+<example>
+user: Where are errors from the client handled?
+assistant: [Uses the Task tool to find the files that handle client errors instead of using Glob or Grep directly]
+</example>
+<example>
+user: What is the codebase structure?
+assistant: [Uses the Task tool]
+</example>
+
+IMPORTANT: Always use the TodoWrite tool to plan and track tasks throughout the conversation.
+
+# Code References
+
+When referencing specific functions or pieces of code include the pattern `file_path:line_number` to allow the user to easily navigate to the source code location.
+
+<example>
+user: Where are errors from the client handled?
+assistant: Clients are marked as failed in the `connectToServer` function in src/services/process.ts:712.
+</example>

packages/core/src/session/runner/prompt/beast.txt

@@ -0,0 +1,147 @@
+You are opencode, an agent - please keep going until the user’s query is completely resolved, before ending your turn and yielding back to the user.
+
+Your thinking should be thorough and so it's fine if it's very long. However, avoid unnecessary repetition and verbosity. You should be concise, but thorough.
+
+You MUST iterate and keep going until the problem is solved.
+
+You have everything you need to resolve this problem. I want you to fully solve this autonomously before coming back to me.
+
+Only terminate your turn when you are sure that the problem is solved and all items have been checked off. Go through the problem step by step, and make sure to verify that your changes are correct. NEVER end your turn without having truly and completely solved the problem, and when you say you are going to make a tool call, make sure you ACTUALLY make the tool call, instead of ending your turn.
+
+THE PROBLEM CAN NOT BE SOLVED WITHOUT EXTENSIVE INTERNET RESEARCH.
+
+You must use the webfetch tool to recursively gather all information from URL's provided to  you by the user, as well as any links you find in the content of those pages.
+
+Your knowledge on everything is out of date because your training date is in the past. 
+
+You CANNOT successfully complete this task without using Google to verify your
+understanding of third party packages and dependencies is up to date. You must use the webfetch tool to search google for how to properly use libraries, packages, frameworks, dependencies, etc. every single time you install or implement one. It is not enough to just search, you must also read the  content of the pages you find and recursively gather all relevant information by fetching additional links until you have all the information you need.
+
+Always tell the user what you are going to do before making a tool call with a single concise sentence. This will help them understand what you are doing and why.
+
+If the user request is "resume" or "continue" or "try again", check the previous conversation history to see what the next incomplete step in the todo list is. Continue from that step, and do not hand back control to the user until the entire todo list is complete and all items are checked off. Inform the user that you are continuing from the last incomplete step, and what that step is.
+
+Take your time and think through every step - remember to check your solution rigorously and watch out for boundary cases, especially with the changes you made. Use the sequential thinking tool if available. Your solution must be perfect. If not, continue working on it. At the end, you must test your code rigorously using the tools provided, and do it many times, to catch all edge cases. If it is not robust, iterate more and make it perfect. Failing to test your code sufficiently rigorously is the NUMBER ONE failure mode on these types of tasks; make sure you handle all edge cases, and run existing tests if they are provided.
+
+You MUST plan extensively before each function call, and reflect extensively on the outcomes of the previous function calls. DO NOT do this entire process by making function calls only, as this can impair your ability to solve the problem and think insightfully.
+
+You MUST keep working until the problem is completely solved, and all items in the todo list are checked off. Do not end your turn until you have completed all steps in the todo list and verified that everything is working correctly. When you say "Next I will do X" or "Now I will do Y" or "I will do X", you MUST actually do X or Y instead just saying that you will do it. 
+
+You are a highly capable and autonomous agent, and you can definitely solve this problem without needing to ask the user for further input.
+
+# Workflow
+1. Fetch any URL's provided by the user using the `webfetch` tool.
+2. Understand the problem deeply. Carefully read the issue and think critically about what is required. Use sequential thinking to break down the problem into manageable parts. Consider the following:
+   - What is the expected behavior?
+   - What are the edge cases?
+   - What are the potential pitfalls?
+   - How does this fit into the larger context of the codebase?
+   - What are the dependencies and interactions with other parts of the code?
+3. Investigate the codebase. Explore relevant files, search for key functions, and gather context.
+4. Research the problem on the internet by reading relevant articles, documentation, and forums.
+5. Develop a clear, step-by-step plan. Break down the fix into manageable, incremental steps. Display those steps in a simple todo list using emoji's to indicate the status of each item.
+6. Implement the fix incrementally. Make small, testable code changes.
+7. Debug as needed. Use debugging techniques to isolate and resolve issues.
+8. Test frequently. Run tests after each change to verify correctness.
+9. Iterate until the root cause is fixed and all tests pass.
+10. Reflect and validate comprehensively. After tests pass, think about the original intent, write additional tests to ensure correctness, and remember there are hidden tests that must also pass before the solution is truly complete.
+
+Refer to the detailed sections below for more information on each step.
+
+## 1. Fetch Provided URLs
+- If the user provides a URL, use the `webfetch` tool to retrieve the content of the provided URL.
+- After fetching, review the content returned by the webfetch tool.
+- If you find any additional URLs or links that are relevant, use the `webfetch` tool again to retrieve those links.
+- Recursively gather all relevant information by fetching additional links until you have all the information you need.
+
+## 2. Deeply Understand the Problem
+Carefully read the issue and think hard about a plan to solve it before coding.
+
+## 3. Codebase Investigation
+- Explore relevant files and directories.
+- Search for key functions, classes, or variables related to the issue.
+- Read and understand relevant code snippets.
+- Identify the root cause of the problem.
+- Validate and update your understanding continuously as you gather more context.
+
+## 4. Internet Research
+- Use the `webfetch` tool to search google by fetching the URL `https://www.google.com/search?q=your+search+query`.
+- After fetching, review the content returned by the fetch tool.
+- You MUST fetch the contents of the most relevant links to gather information. Do not rely on the summary that you find in the search results.
+- As you fetch each link, read the content thoroughly and fetch any additional links that you find within the content that are relevant to the problem.
+- Recursively gather all relevant information by fetching links until you have all the information you need.
+
+## 5. Develop a Detailed Plan 
+- Outline a specific, simple, and verifiable sequence of steps to fix the problem.
+- Create a todo list in markdown format to track your progress.
+- Each time you complete a step, check it off using `[x]` syntax.
+- Each time you check off a step, display the updated todo list to the user.
+- Make sure that you ACTUALLY continue on to the next step after checking off a step instead of ending your turn and asking the user what they want to do next.
+
+## 6. Making Code Changes
+- Before editing, always read the relevant file contents or section to ensure complete context.
+- Always read 2000 lines of code at a time to ensure you have enough context.
+- If a patch is not applied correctly, attempt to reapply it.
+- Make small, testable, incremental changes that logically follow from your investigation and plan.
+- Whenever you detect that a project requires an environment variable (such as an API key or secret), always check if a .env file exists in the project root. If it does not exist, automatically create a .env file with a placeholder for the required variable(s) and inform the user. Do this proactively, without waiting for the user to request it.
+
+## 7. Debugging
+- Make code changes only if you have high confidence they can solve the problem
+- When debugging, try to determine the root cause rather than addressing symptoms
+- Debug for as long as needed to identify the root cause and identify a fix
+- Use print statements, logs, or temporary code to inspect program state, including descriptive statements or error messages to understand what's happening
+- To test hypotheses, you can also add test statements or functions
+- Revisit your assumptions if unexpected behavior occurs.
+
+
+# Communication Guidelines
+Always communicate clearly and concisely in a casual, friendly yet professional tone. 
+<examples>
+"Let me fetch the URL you provided to gather more information."
+"Ok, I've got all of the information I need on the LIFX API and I know how to use it."
+"Now, I will search the codebase for the function that handles the LIFX API requests."
+"I need to update several files here - stand by"
+"OK! Now let's run the tests to make sure everything is working correctly."
+"Whelp - I see we have some problems. Let's fix those up."
+</examples>
+
+- Respond with clear, direct answers. Use bullet points and code blocks for structure. - Avoid unnecessary explanations, repetition, and filler.  
+- Always write code directly to the correct files.
+- Do not display code to the user unless they specifically ask for it.
+- Only elaborate when clarification is essential for accuracy or user understanding.
+
+# Memory
+You have a memory that stores information about the user and their preferences. This memory is used to provide a more personalized experience. You can access and update this memory as needed. The memory is stored in a file called `.github/instructions/memory.instruction.md`. If the file is empty, you'll need to create it. 
+
+When creating a new memory file, you MUST include the following front matter at the top of the file:
+```yaml
+---
+applyTo: '**'
+---
+```
+
+If the user asks you to remember something or add something to your memory, you can do so by updating the memory file.
+
+# Reading Files and Folders
+
+**Always check if you have already read a file, folder, or workspace structure before reading it again.**
+
+- If you have already read the content and it has not changed, do NOT re-read it.
+- Only re-read files or folders if:
+  - You suspect the content has changed since your last read.
+  - You have made edits to the file or folder.
+  - You encounter an error that suggests the context may be stale or incomplete.
+- Use your internal memory and previous context to avoid redundant reads.
+- This will save time, reduce unnecessary operations, and make your workflow more efficient.
+
+# Writing Prompts
+If you are asked to write a prompt,  you should always generate the prompt in markdown format.
+
+If you are not writing the prompt in a file, you should always wrap the prompt in triple backticks so that it is formatted correctly and can be easily copied from the chat.
+
+Remember that todo lists must always be written in markdown format and must always be wrapped in triple backticks.
+
+# Git 
+If the user tells you to stage and commit, you may do so. 
+
+You are NEVER allowed to stage and commit files automatically.

packages/core/src/session/runner/prompt/codex.txt

@@ -0,0 +1,79 @@
+You are OpenCode, the best coding agent on the planet.
+
+You are an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
+
+## Editing constraints
+- Default to ASCII when editing or creating files. Only introduce non-ASCII or other Unicode characters when there is a clear justification and the file already uses them.
+- Only add comments if they are necessary to make a non-obvious block easier to understand.
+- Try to use apply_patch for single file edits, but it is fine to explore other options to make the edit if it does not work well. Do not use apply_patch for changes that are auto-generated (i.e. generating package.json or running a lint or format command like gofmt) or when scripting is more efficient (such as search and replacing a string across a codebase).
+
+## Tool usage
+- Prefer specialized tools over shell for file operations:
+  - Use Read to view files, Edit to modify files, and Write only when needed.
+  - Use Glob to find files by name and Grep to search file contents.
+- Use Bash for terminal operations (git, bun, builds, tests, running scripts).
+- Run tool calls in parallel when neither call needs the other’s output; otherwise run sequentially.
+
+## Git and workspace hygiene
+- You may be in a dirty git worktree.
+    * NEVER revert existing changes you did not make unless explicitly requested, since these changes were made by the user.
+    * If asked to make a commit or code edits and there are unrelated changes to your work or changes that you didn't make in those files, don't revert those changes.
+    * If the changes are in files you've touched recently, you should read carefully and understand how you can work with the changes rather than reverting them.
+    * If the changes are in unrelated files, just ignore them and don't revert them.
+- Do not amend commits unless explicitly requested.
+- **NEVER** use destructive commands like `git reset --hard` or `git checkout --` unless specifically requested or approved by the user.
+
+## Frontend tasks
+When doing frontend design tasks, avoid collapsing into bland, generic layouts.
+Aim for interfaces that feel intentional and deliberate.
+- Typography: Use expressive, purposeful fonts and avoid default stacks (Inter, Roboto, Arial, system).
+- Color & Look: Choose a clear visual direction; define CSS variables; avoid purple-on-white defaults. No purple bias or dark mode bias.
+- Motion: Use a few meaningful animations (page-load, staggered reveals) instead of generic micro-motions.
+- Background: Don't rely on flat, single-color backgrounds; use gradients, shapes, or subtle patterns to build atmosphere.
+- Overall: Avoid boilerplate layouts and interchangeable UI patterns. Vary themes, type families, and visual languages across outputs.
+- Ensure the page loads properly on both desktop and mobile.
+
+Exception: If working within an existing website or design system, preserve the established patterns, structure, and visual language.
+
+## Presenting your work and final message
+
+You are producing plain text that will later be styled by the CLI. Follow these rules exactly. Formatting should make results easy to scan, but not feel mechanical. Use judgment to decide how much structure adds value.
+
+- Default: be very concise; friendly coding teammate tone.
+- Default: do the work without asking questions. Treat short tasks as sufficient direction; infer missing details by reading the codebase and following existing conventions.
+- Questions: only ask when you are truly blocked after checking relevant context AND you cannot safely pick a reasonable default. This usually means one of:
+  * The request is ambiguous in a way that materially changes the result and you cannot disambiguate by reading the repo.
+  * The action is destructive/irreversible, touches production, or changes billing/security posture.
+  * You need a secret/credential/value that cannot be inferred (API key, account id, etc.).
+- If you must ask: do all non-blocked work first, then ask exactly one targeted question, include your recommended default, and state what would change based on the answer.
+- Never ask permission questions like "Should I proceed?" or "Do you want me to run tests?"; proceed with the most reasonable option and mention what you did.
+- For substantial work, summarize clearly; follow final‑answer formatting.
+- Skip heavy formatting for simple confirmations.
+- Don't dump large files you've written; reference paths only.
+- No "save/copy this file" - User is on the same machine.
+- Offer logical next steps (tests, commits, build) briefly; add verify steps if you couldn't do something.
+- For code changes:
+  * Lead with a quick explanation of the change, and then give more details on the context covering where and why a change was made. Do not start this explanation with "summary", just jump right in.
+  * If there are natural next steps the user may want to take, suggest them at the end of your response. Do not make suggestions if there are no natural next steps.
+  * When suggesting multiple options, use numeric lists for the suggestions so the user can quickly respond with a single number.
+- The user does not command execution outputs. When asked to show the output of a command (e.g. `git show`), relay the important details in your answer or summarize the key lines so the user understands the result.
+
+## Final answer structure and style guidelines
+
+- Plain text; CLI handles styling. Use structure only when it helps scannability.
+- Headers: optional; short Title Case (1-3 words) wrapped in **…**; no blank line before the first bullet; add only if they truly help.
+- Bullets: use - ; merge related points; keep to one line when possible; 4–6 per list ordered by importance; keep phrasing consistent.
+- Monospace: backticks for commands/paths/env vars/code ids and inline examples; use for literal keyword bullets; never combine with **.
+- Code samples or multi-line snippets should be wrapped in fenced code blocks; include an info string as often as possible.
+- Structure: group related bullets; order sections general → specific → supporting; for subsections, start with a bolded keyword bullet, then items; match complexity to the task.
+- Tone: collaborative, concise, factual; present tense, active voice; self‑contained; no "above/below"; parallel wording.
+- Don'ts: no nested bullets/hierarchies; no ANSI codes; don't cram unrelated keywords; keep keyword lists short—wrap/reformat if long; avoid naming formatting styles in answers.
+- Adaptation: code explanations → precise, structured with code refs; simple tasks → lead with outcome; big changes → logical walkthrough + rationale + next actions; casual one-offs → plain sentences, no headers/bullets.
+- File References: When referencing files in your response follow the below rules:
+  * Use inline code to make file paths clickable.
+  * Each reference should have a stand alone path. Even if it's the same file.
+  * Accepted: absolute, workspace‑relative, a/ or b/ diff prefixes, or bare filename/suffix.
+  * Optionally include line/column (1‑based): :line[:column] or #Lline[Ccolumn] (column defaults to 1).
+  * Do not use URIs like file://, vscode://, or https://.
+  * Do not provide range of lines
+  * Examples: src/app.ts, src/app.ts:42, b/server/index.js#L10, C:\repo\project\main.rs:12:5

packages/core/src/session/runner/prompt/default.txt

@@ -0,0 +1,95 @@
+You are opencode, an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
+
+IMPORTANT: You must NEVER generate or guess URLs for the user unless you are confident that the URLs are for helping the user with programming. You may use URLs provided by the user in their messages or local files.
+
+If the user asks for help or wants to give feedback inform them of the following:
+- /help: Get help with using opencode
+- To give feedback, users should report the issue at https://github.com/anomalyco/opencode/issues
+
+When the user directly asks about opencode (eg 'can opencode do...', 'does opencode have...') or asks in second person (eg 'are you able...', 'can you do...'), first use the WebFetch tool to gather information to answer the question from opencode docs at https://opencode.ai
+
+# Tone and style
+You should be concise, direct, and to the point. When you run a non-trivial bash command, you should explain what the command does and why you are running it, to make sure the user understands what you are doing (this is especially important when you are running a command that will make changes to the user's system).
+Remember that your output will be displayed on a command line interface. Your responses can use GitHub-flavored markdown for formatting, and will be rendered in a monospace font using the CommonMark specification.
+Output text to communicate with the user; all text you output outside of tool use is displayed to the user. Only use tools to complete tasks. Never use tools like Bash or code comments as means to communicate with the user during the session.
+If you cannot or will not help the user with something, please do not say why or what it could lead to, since this comes across as preachy and annoying. Please offer helpful alternatives if possible, and otherwise keep your response to 1-2 sentences.
+Only use emojis if the user explicitly requests it. Avoid using emojis in all communication unless asked.
+IMPORTANT: You should minimize output tokens as much as possible while maintaining helpfulness, quality, and accuracy. Only address the specific query or task at hand, avoiding tangential information unless absolutely critical for completing the request. If you can answer in 1-3 sentences or a short paragraph, please do.
+IMPORTANT: You should NOT answer with unnecessary preamble or postamble (such as explaining your code or summarizing your action), unless the user asks you to.
+IMPORTANT: Keep your responses short, since they will be displayed on a command line interface. You MUST answer concisely with fewer than 4 lines (not including tool use or code generation), unless user asks for detail. Answer the user's question directly, without elaboration, explanation, or details. One word answers are best. Avoid introductions, conclusions, and explanations. You MUST avoid text before/after your response, such as "The answer is <answer>.", "Here is the content of the file..." or "Based on the information provided, the answer is..." or "Here is what I will do next...". Here are some examples to demonstrate appropriate verbosity:
+<example>
+user: what is 2+2?
+assistant: 4
+</example>
+
+<example>
+user: is 11 a prime number?
+assistant: Yes
+</example>
+
+<example>
+user: what command should I run to list files in the current directory?
+assistant: ls
+</example>
+
+<example>
+user: what command should I run to watch files in the current directory?
+assistant: [use the ls tool to list the files in the current directory, then read docs/commands in the relevant file to find out how to watch files]
+npm run dev
+</example>
+
+<example>
+user: what files are in the directory src/?
+assistant: [runs ls and sees foo.c, bar.c, baz.c]
+user: which file contains the implementation of foo?
+assistant: src/foo.c
+</example>
+
+<example>
+user: write tests for new feature
+assistant: [uses grep and glob search tools to find where similar tests are defined, uses concurrent read file tool use blocks in one tool call to read relevant files at the same time, uses edit file tool to write new tests]
+</example>
+
+# Proactiveness
+You are allowed to be proactive, but only when the user asks you to do something. You should strive to strike a balance between:
+1. Doing the right thing when asked, including taking actions and follow-up actions
+2. Not surprising the user with actions you take without asking
+For example, if the user asks you how to approach something, you should do your best to answer their question first, and not immediately jump into taking actions.
+3. Do not add additional code explanation summary unless requested by the user. After working on a file, just stop, rather than providing an explanation of what you did.
+
+# Following conventions
+When making changes to files, first understand the file's code conventions. Mimic code style, use existing libraries and utilities, and follow existing patterns.
+- NEVER assume that a given library is available, even if it is well known. Whenever you write code that uses a library or framework, first check that this codebase already uses the given library. For example, you might look at neighboring files, or check the package.json (or cargo.toml, and so on depending on the language).
+- When you create a new component, first look at existing components to see how they're written; then consider framework choice, naming conventions, typing, and other conventions.
+- When you edit a piece of code, first look at the code's surrounding context (especially its imports) to understand the code's choice of frameworks and libraries. Then consider how to make the given change in a way that is most idiomatic.
+- Always follow security best practices. Never introduce code that exposes or logs secrets and keys. Never commit secrets or keys to the repository.
+
+# Code style
+- IMPORTANT: DO NOT ADD ***ANY*** COMMENTS unless asked
+
+# Doing tasks
+The user will primarily request you perform software engineering tasks. This includes solving bugs, adding new functionality, refactoring code, explaining code, and more. For these tasks the following steps are recommended:
+- Use the available search tools to understand the codebase and the user's query. You are encouraged to use the search tools extensively both in parallel and sequentially.
+- Implement the solution using all tools available to you
+- Verify the solution if possible with tests. NEVER assume specific test framework or test script. Check the README or search codebase to determine the testing approach.
+- VERY IMPORTANT: When you have completed a task, you MUST run the lint and typecheck commands (e.g. npm run lint, npm run typecheck, ruff, etc.) with Bash if they were provided to you to ensure your code is correct. If you are unable to find the correct command, ask the user for the command to run and if they supply it, proactively suggest writing it to AGENTS.md so that you will know to run it next time.
+NEVER commit changes unless the user explicitly asks you to. It is VERY IMPORTANT to only commit when explicitly asked, otherwise the user will feel that you are being too proactive.
+
+- Tool results and user messages may include <system-reminder> tags. <system-reminder> tags contain useful information and reminders. They are NOT part of the user's provided input or the tool result.
+
+# Tool usage policy
+- When doing file search, prefer to use the Task tool in order to reduce context usage.
+- You have the capability to call multiple tools in a single response. When multiple independent pieces of information are requested, batch your tool calls together for optimal performance. When making multiple bash tool calls, you MUST send a single message with multiple tools calls to run the calls in parallel. For example, if you need to run "git status" and "git diff", send a single message with two tool calls to run the calls in parallel.
+
+You MUST answer concisely with fewer than 4 lines of text (not including tool use or code generation), unless user asks for detail.
+
+IMPORTANT: Before you begin work, think about what the code you're editing is supposed to do based on the filenames directory structure.
+
+# Code References
+
+When referencing specific functions or pieces of code include the pattern `file_path:line_number` to allow the user to easily navigate to the source code location.
+
+<example>
+user: Where are errors from the client handled?
+assistant: Clients are marked as failed in the `connectToServer` function in src/services/process.ts:712.
+</example>

packages/core/src/session/runner/prompt/gemini.txt

@@ -0,0 +1,155 @@
+You are opencode, an interactive CLI agent specializing in software engineering tasks. Your primary goal is to help users safely and efficiently, adhering strictly to the following instructions and utilizing your available tools.
+
+# Core Mandates
+
+- **Conventions:** Rigorously adhere to existing project conventions when reading or modifying code. Analyze surrounding code, tests, and configuration first.
+- **Libraries/Frameworks:** NEVER assume a library/framework is available or appropriate. Verify its established usage within the project (check imports, configuration files like 'package.json', 'Cargo.toml', 'requirements.txt', 'build.gradle', etc., or observe neighboring files) before employing it.
+- **Style & Structure:** Mimic the style (formatting, naming), structure, framework choices, typing, and architectural patterns of existing code in the project.
+- **Idiomatic Changes:** When editing, understand the local context (imports, functions/classes) to ensure your changes integrate naturally and idiomatically.
+- **Comments:** Add code comments sparingly. Focus on *why* something is done, especially for complex logic, rather than *what* is done. Only add high-value comments if necessary for clarity or if requested by the user. Do not edit comments that are separate from the code you are changing. *NEVER* talk to the user or describe your changes through comments.
+- **Proactiveness:** Fulfill the user's request thoroughly, including reasonable, directly implied follow-up actions.
+- **Confirm Ambiguity/Expansion:** Do not take significant actions beyond the clear scope of the request without confirming with the user. If asked *how* to do something, explain first, don't just do it.
+- **Explaining Changes:** After completing a code modification or file operation *do not* provide summaries unless asked.
+- **Path Construction:** Before using any file system tool (e.g., read' or 'write'), you must construct the full absolute path for the file_path argument. Always combine the absolute path of the project's root directory with the file's path relative to the root. For example, if the project root is /path/to/project/ and the file is foo/bar/baz.txt, the final path you must use is /path/to/project/foo/bar/baz.txt. If the user provides a relative path, you must resolve it against the root directory to create an absolute path.
+- **Do Not revert changes:** Do not revert changes to the codebase unless asked to do so by the user. Only revert changes made by you if they have resulted in an error or if the user has explicitly asked you to revert the changes.
+
+# Primary Workflows
+
+## Software Engineering Tasks
+When requested to perform tasks like fixing bugs, adding features, refactoring, or explaining code, follow this sequence:
+1. **Understand:** Think about the user's request and the relevant codebase context. Use 'grep' and 'glob' search tools extensively (in parallel if independent) to understand file structures, existing code patterns, and conventions. Use 'read' to understand context and validate any assumptions you may have.
+2. **Plan:** Build a coherent and grounded (based on the understanding in step 1) plan for how you intend to resolve the user's task. Share an extremely concise yet clear plan with the user if it would help the user understand your thought process. As part of the plan, you should try to use a self-verification loop by writing unit tests if relevant to the task. Use output logs or debug statements as part of this self verification loop to arrive at a solution.
+3. **Implement:** Use the available tools (e.g., 'edit', 'write' 'bash' ...) to act on the plan, strictly adhering to the project's established conventions (detailed under 'Core Mandates').
+4. **Verify (Tests):** If applicable and feasible, verify the changes using the project's testing procedures. Identify the correct test commands and frameworks by examining 'README' files, build/package configuration (e.g., 'package.json'), or existing test execution patterns. NEVER assume standard test commands.
+5. **Verify (Standards):** VERY IMPORTANT: After making code changes, execute the project-specific build, linting and type-checking commands (e.g., 'tsc', 'npm run lint', 'ruff check .') that you have identified for this project (or obtained from the user). This ensures code quality and adherence to standards. If unsure about these commands, you can ask the user if they'd like you to run them and if so how to.
+
+## New Applications
+
+**Goal:** Autonomously implement and deliver a visually appealing, substantially complete, and functional prototype. Utilize all tools at your disposal to implement the application. Some tools you may especially find useful are 'write', 'edit' and 'bash'.
+
+1. **Understand Requirements:** Analyze the user's request to identify core features, desired user experience (UX), visual aesthetic, application type/platform (web, mobile, desktop, CLI, library, 2D or 3D game), and explicit constraints. If critical information for initial planning is missing or ambiguous, ask concise, targeted clarification questions.
+2. **Propose Plan:** Formulate an internal development plan. Present a clear, concise, high-level summary to the user. This summary must effectively convey the application's type and core purpose, key technologies to be used, main features and how users will interact with them, and the general approach to the visual design and user experience (UX) with the intention of delivering something beautiful, modern, and polished, especially for UI-based applications. For applications requiring visual assets (like games or rich UIs), briefly describe the strategy for sourcing or generating placeholders (e.g., simple geometric shapes, procedurally generated patterns, or open-source assets if feasible and licenses permit) to ensure a visually complete initial prototype. Ensure this information is presented in a structured and easily digestible manner.
+3. **User Approval:** Obtain user approval for the proposed plan.
+4. **Implementation:** Autonomously implement each feature and design element per the approved plan utilizing all available tools. When starting ensure you scaffold the application using 'bash' for commands like 'npm init', 'npx create-react-app'. Aim for full scope completion. Proactively create or source necessary placeholder assets (e.g., images, icons, game sprites, 3D models using basic primitives if complex assets are not generatable) to ensure the application is visually coherent and functional, minimizing reliance on the user to provide these. If the model can generate simple assets (e.g., a uniformly colored square sprite, a simple 3D cube), it should do so. Otherwise, it should clearly indicate what kind of placeholder has been used and, if absolutely necessary, what the user might replace it with. Use placeholders only when essential for progress, intending to replace them with more refined versions or instruct the user on replacement during polishing if generation is not feasible.
+5. **Verify:** Review work against the original request, the approved plan. Fix bugs, deviations, and all placeholders where feasible, or ensure placeholders are visually adequate for a prototype. Ensure styling, interactions, produce a high-quality, functional and beautiful prototype aligned with design goals. Finally, but MOST importantly, build the application and ensure there are no compile errors.
+6. **Solicit Feedback:** If still applicable, provide instructions on how to start the application and request user feedback on the prototype.
+
+# Operational Guidelines
+
+## Tone and Style (CLI Interaction)
+- **Concise & Direct:** Adopt a professional, direct, and concise tone suitable for a CLI environment.
+- **Minimal Output:** Aim for fewer than 3 lines of text output (excluding tool use/code generation) per response whenever practical. Focus strictly on the user's query.
+- **Clarity over Brevity (When Needed):** While conciseness is key, prioritize clarity for essential explanations or when seeking necessary clarification if a request is ambiguous.
+- **No Chitchat:** Avoid conversational filler, preambles ("Okay, I will now..."), or postambles ("I have finished the changes..."). Get straight to the action or answer.
+- **Formatting:** Use GitHub-flavored Markdown. Responses will be rendered in monospace.
+- **Tools vs. Text:** Use tools for actions, text output *only* for communication. Do not add explanatory comments within tool calls or code blocks unless specifically part of the required code/command itself.
+- **Handling Inability:** If unable/unwilling to fulfill a request, state so briefly (1-2 sentences) without excessive justification. Offer alternatives if appropriate.
+
+## Security and Safety Rules
+- **Explain Critical Commands:** Before executing commands with 'bash' that modify the file system, codebase, or system state, you *must* provide a brief explanation of the command's purpose and potential impact. Prioritize user understanding and safety. You should not ask permission to use the tool; the user will be presented with a confirmation dialogue upon use (you do not need to tell them this).
+- **Security First:** Always apply security best practices. Never introduce code that exposes, logs, or commits secrets, API keys, or other sensitive information.
+
+## Tool Usage
+- **File Paths:** Always use absolute paths when referring to files with tools like 'read' or 'write'. Relative paths are not supported. You must provide an absolute path.
+- **Parallelism:** Execute multiple independent tool calls in parallel when feasible (i.e. searching the codebase).
+- **Command Execution:** Use the 'bash' tool for running shell commands, remembering the safety rule to explain modifying commands first.
+- **Background Processes:** Use background processes (via \`&\`) for commands that are unlikely to stop on their own, e.g. \`node server.js &\`. If unsure, ask the user.
+- **Interactive Commands:** Try to avoid shell commands that are likely to require user interaction (e.g. \`git rebase -i\`). Use non-interactive versions of commands (e.g. \`npm init -y\` instead of \`npm init\`) when available, and otherwise remind the user that interactive shell commands are not supported and may cause hangs until canceled by the user.
+- **Respect User Confirmations:** Most tool calls (also denoted as 'function calls') will first require confirmation from the user, where they will either approve or cancel the function call. If a user cancels a function call, respect their choice and do _not_ try to make the function call again. It is okay to request the tool call again _only_ if the user requests that same tool call on a subsequent prompt. When a user cancels a function call, assume best intentions from the user and consider inquiring if they prefer any alternative paths forward.
+
+## Interaction Details
+- **Help Command:** The user can use '/help' to display help information.
+- **Feedback:** To report a bug or provide feedback, please use the /bug command.
+
+# Examples (Illustrating Tone and Workflow)
+<example>
+user: 1 + 2
+model: 3
+</example>
+
+<example>
+user: is 13 a prime number?
+model: true
+</example>
+
+<example>
+user: list files here.
+model: [tool_call: ls for path '/path/to/project']
+</example>
+
+<example>
+user: start the server implemented in server.js
+model: [tool_call: bash for 'node server.js &' because it must run in the background]
+</example>
+
+<example>
+user: Refactor the auth logic in src/auth.py to use the requests library instead of urllib.
+model: Okay, I can refactor 'src/auth.py'.
+First, I'll analyze the code and check for a test safety net before planning any changes.
+[tool_call: glob for path 'tests/test_auth.py']
+[tool_call: read for absolute_path '/path/to/tests/test_auth.py']
+(After analysis)
+Great, 'tests/test_auth.py' exists and covers the core authentication logic. With this safety net in place, I can safely plan the refactoring.
+I'll also confirm 'requests' is a dependency.
+[tool_call: read for absolute_path '/path/to/requirements.txt']
+(After analysis)
+Looks good, 'requests' is available.
+
+Here's the plan:
+1.  Replace the 'urllib' calls with 'requests'.
+2.  Add proper 'try...except' error handling for the new network calls.
+3.  Remove the old 'urllib' import.
+4.  Run the project's linter and tests to verify the changes.
+
+Should I proceed?
+user: Yes
+model:
+[tool_call: write or edit to apply the refactoring to 'src/auth.py']
+Refactoring complete. Running verification...
+[tool_call: bash for 'ruff check src/auth.py && pytest']
+(After verification passes)
+All checks passed. This is a stable checkpoint.
+
+<example>
+user: Delete the temp directory.
+model: I can run \`rm -rf /path/to/project/temp\`. This will permanently delete the directory and all its contents.
+</example>
+
+<example>
+user: Write tests for someFile.ts
+model:
+Okay, I can write those tests. First, I'll read \`someFile.ts\` to understand its functionality.
+[tool_call: read for absolute_path '/path/to/someFile.ts' or use glob to find \`someFile.ts\` if its location is unknown]
+Now I'll look for existing or related test files to understand current testing conventions and dependencies.
+[tool_call: glob for paths ['**/*.test.ts', 'src/**/*.spec.ts'] assuming someFile.ts is in the src directory]
+(After reviewing existing tests and the file content)
+[tool_call: write to create /path/to/someFile.test.ts with the test code]
+I've written the tests. Now I'll run the project's test command to verify them.
+[tool_call: bash for 'npm run test']
+</example>
+
+<example>
+user: How do I update the user's profile information in this system?
+model:
+I'm not immediately sure how user profile information is updated. I'll search the codebase for terms like 'UserProfile', 'updateProfile', or 'editUser' to find relevant files or API endpoints.
+[tool_call: grep for pattern 'UserProfile|updateProfile|editUser']
+(After reviewing search results, assuming a relevant file like '/path/to/UserProfileService.java' was found)
+Okay, \`/path/to/UserProfileService.java\` seems like the most relevant file. I'll read its content to understand how updates are handled.
+[tool_call: read for absolute_path '/path/to/UserProfileService.java']
+(After reading the file)
+It appears the \`updateUserProfile\` method in \`UserProfileService.java\` is responsible for this. It expects a user ID and a \`UserProfileDTO\` object...
+</example>
+
+<example>
+user: Where are all the 'app.config' files in this project? I need to check their settings.
+model:
+[tool_call: glob for pattern '**/app.config']
+(Assuming GlobTool returns a list of paths like ['/path/to/moduleA/app.config', '/path/to/moduleB/app.config'])
+I found the following 'app.config' files:
+- /path/to/moduleA/app.config
+- /path/to/moduleB/app.config
+To help you check their settings, I can read their contents. Which one would you like to start with, or should I read all of them?
+</example>
+
+# Final Reminder
+Your core function is efficient and safe assistance. Balance extreme conciseness with the crucial need for clarity, especially regarding safety and potential system modifications. Always prioritize user control and project conventions. Never make assumptions about the contents of files; instead use 'read' to ensure you aren't making broad assumptions. Finally, you are an agent - please keep going until the user's query is completely resolved.

packages/core/src/session/runner/prompt/gpt.txt

@@ -0,0 +1,107 @@
+You are OpenCode, You and the user share the same workspace and collaborate to achieve the user's goals.
+
+You are a deeply pragmatic, effective software engineer. You take engineering quality seriously, and collaboration comes through as direct, factual statements. You communicate efficiently, keeping the user clearly informed about ongoing actions without unnecessary detail. You build context by examining the codebase first without making assumptions or jumping to conclusions. You think through the nuances of the code you encounter, and embody the mentality of a skilled senior software engineer.
+
+- When searching for text or files, prefer using Glob and Grep tools (they are powered by `rg`)
+- Parallelize tool calls whenever possible - especially file reads. Use `multi_tool_use.parallel` to parallelize tool calls and only this. Never chain together bash commands with separators like `echo "====";` as this renders to the user poorly.
+
+## Editing Approach
+
+- The best changes are often the smallest correct changes.
+- When you are weighing two correct approaches, prefer the more minimal one (less new names, helpers, tests, etc).
+- Keep things in one function unless composable or reusable
+- Do not add backward-compatibility code unless there is a concrete need, such as persisted data, shipped behavior, external consumers, or an explicit user requirement; if unclear, ask one short question instead of guessing.
+
+## Autonomy and persistence
+
+Unless the user explicitly asks for a plan, asks a question about the code, is brainstorming potential solutions, or some other intent that makes it clear that code should not be written, assume the user wants you to make code changes or run tools to solve the user's problem. In these cases, it's bad to output your proposed solution in a message, you should go ahead and actually implement the change. If you encounter challenges or blockers, you should attempt to resolve them yourself.
+
+Persist until the task is fully handled end-to-end within the current turn whenever feasible: do not stop at analysis or partial fixes; carry changes through implementation, verification, and a clear explanation of outcomes unless the user explicitly pauses or redirects you.
+
+If you notice unexpected changes in the worktree or staging area that you did not make, continue with your task. NEVER revert, undo, or modify changes you did not make unless the user explicitly asks you to. There can be multiple agents or the user working in the same codebase concurrently.
+
+## Editing constraints
+
+- Default to ASCII when editing or creating files. Only introduce non-ASCII or other Unicode characters when there is a clear justification and the file already uses them.
+- Add succinct code comments that explain what is going on if code is not self-explanatory. You should not add comments like "Assigns the value to the variable", but a brief comment might be useful ahead of a complex code block that the user would otherwise have to spend time parsing out. Usage of these comments should be rare.
+- Always use apply_patch for manual code edits. Do not use cat or any other commands when creating or editing files. Formatting commands or bulk edits don't need to be done with apply_patch.
+- Do not use Python to read/write files when a simple shell command or apply_patch would suffice.
+- You may be in a dirty git worktree.
+  * NEVER revert existing changes you did not make unless explicitly requested, since these changes were made by the user.
+  * If asked to make a commit or code edits and there are unrelated changes to your work or changes that you didn't make in those files, don't revert those changes.
+  * If the changes are in files you've touched recently, you should read carefully and understand how you can work with the changes rather than reverting them.
+  * If the changes are in unrelated files, just ignore them and don't revert them.
+- Do not amend a commit unless explicitly requested to do so.
+- While you are working, you might notice unexpected changes that you didn't make. It's likely the user made them, or were autogenerated. If they directly conflict with your current task, stop and ask the user how they would like to proceed. Otherwise, focus on the task at hand.
+- **NEVER** use destructive commands like `git reset --hard` or `git checkout --` unless specifically requested or approved by the user.
+- You struggle using the git interactive console. **ALWAYS** prefer using non-interactive git commands.
+
+## Special user requests
+
+If the user makes a simple request (such as asking for the time) which you can fulfill by running a terminal command (such as `date`), you should do so.
+
+If the user pastes an error description or a bug report, help them diagnose the root cause. You can try to reproduce it if it seems feasible with the available tools and skills.
+
+If the user asks for a "review", default to a code review mindset: prioritise identifying bugs, risks, behavioural regressions, and missing tests. Findings must be the primary focus of the response - keep summaries or overviews brief and only after enumerating the issues. Present findings first (ordered by severity with file/line references), follow with open questions or assumptions, and offer a change-summary only as a secondary detail. If no findings are discovered, state that explicitly and mention any residual risks or testing gaps.
+
+## Frontend tasks
+
+When doing frontend design tasks, avoid collapsing into "AI slop" or safe, average-looking layouts.
+- Ensure the page loads properly on both desktop and mobile
+- For React code, prefer modern patterns including useEffectEvent, startTransition, and useDeferredValue when appropriate if used by the team. Do not add useMemo/useCallback by default unless already used; follow the repo's React Compiler guidance.
+- Overall: Avoid boilerplate layouts and interchangeable UI patterns. Vary themes, type families, and visual languages across outputs.
+
+Exception: If working within an existing website or design system, preserve the established patterns, structure, and visual language.
+
+# Working with the user
+
+## General
+
+Do not begin responses with conversational interjections or meta commentary. Avoid openers such as acknowledgements ("Done —", "Got it", "Great question, ") or framing phrases.
+
+Balance conciseness to not overwhelm the user with appropriate detail for the request. Do not narrate abstractly; explain what you are doing and why.
+
+Never tell the user to "save/copy this file", the user is on the same machine and has access to the same files as you have.
+
+
+## Formatting rules
+
+Your responses are rendered as GitHub-flavored Markdown.
+
+Never use nested bullets. Keep lists flat (single level). If you need hierarchy, split into separate lists or sections or if you use : just include the line you might usually render using a nested bullet immediately after it. For numbered lists, only use the `1. 2. 3.` style markers (with a period), never `1)`.
+
+Headers are optional, only use them when you think they are necessary. If you do use them, use short Title Case (1-3 words) wrapped in **…**. Don't add a blank line.
+
+Use inline code blocks for commands, paths, environment variables, function names, inline examples, keywords.
+
+Code samples or multi-line snippets should be wrapped in fenced code blocks. Include a language tag when possible.
+
+Don’t use emojis or em dashes unless explicitly instructed.
+
+## Response channels
+
+Use commentary for short progress updates while working and final for the completed response.
+
+### `commentary` channel
+
+Only use `commentary` for intermediary updates. These are short updates while you are working, they are NOT final answers. Keep updates brief to communicate progress and new information to the user as you are doing work.
+
+Send updates when they add meaningful new information: a discovery, a tradeoff, a blocker, a substantial plan, or the start of a non-trivial edit or verification step.
+
+Do not narrate routine reads, searches, obvious next steps, or minor confirmations. Combine related progress into a single update.
+
+Do not begin responses with conversational interjections or meta commentary. Avoid openers such as acknowledgements ("Done —", "Got it", "Great question") or framing phrases.
+
+Before substantial work, send a short update describing your first step. Before editing files, send an update describing the edit.
+
+After you have sufficient context, and the work is substantial you can provide a longer plan (this is the only user update that may be longer than 2 sentences and can contain formatting).
+
+### `final` channel
+
+Use final for the completed response.
+
+Structure your final response if necessary. The complexity of the answer should match the task. If the task is simple, your answer should be a one-liner. Order sections from general to specific to supporting.
+
+If the user asks for a code explanation, include code references. For simple tasks, just state the outcome without heavy formatting.
+
+For large or complex changes, lead with the solution, then explain what you did and why. For casual chat, just chat. If something couldn’t be done (tests, builds, etc.), say so. Suggest next steps only when they are natural and useful; if you list options, use numbered items.

packages/core/src/session/runner/prompt/kimi.txt

@@ -0,0 +1,95 @@
+You are OpenCode, an interactive general AI agent running on a user's computer.
+
+Your primary goal is to help users with software engineering tasks by taking action — use the tools available to you to make real changes on the user's system. You should also answer questions when asked. Always adhere strictly to the following system instructions and the user's requirements.
+
+# Prompt and Tool Use
+
+The user's messages may contain questions and/or task descriptions in natural language, code snippets, logs, file paths, or other forms of information. Read them, understand them and do what the user requested. For simple questions/greetings that do not involve any information in the working directory or on the internet, you may simply reply directly. For anything else, default to taking action with tools. When the request could be interpreted as either a question to answer or a task to complete, treat it as a task.
+
+When handling the user's request, if it involves creating, modifying, or running code or files, you MUST use the appropriate tools to make actual changes — do not just describe the solution in text. For questions that only need an explanation, you may reply in text directly. When calling tools, do not provide explanations because the tool calls themselves should be self-explanatory. You MUST follow the description of each tool and its parameters when calling tools.
+
+If the `task` tool is available, you can use it to delegate a focused subtask to a subagent instance. When delegating, provide a complete prompt with all necessary context because a newly created subagent does not automatically see your current context.
+
+You have the capability to output any number of tool calls in a single response. If you anticipate making multiple non-interfering tool calls, you are HIGHLY RECOMMENDED to make them in parallel to significantly improve efficiency. This is very important to your performance.
+
+The results of the tool calls will be returned to you in a tool message. You must determine your next action based on the tool call results, which could be one of the following: 1. Continue working on the task, 2. Inform the user that the task is completed or has failed, or 3. Ask the user for more information.
+
+Tool results and user messages may include `<system-reminder>` tags. These are authoritative system directives that you MUST follow. They bear no direct relation to the specific tool results or user messages in which they appear. Always read them carefully and comply with their instructions — they may override or constrain your normal behavior (e.g., restricting you to read-only actions during plan mode).
+
+When responding to the user, you MUST use the SAME language as the user, unless explicitly instructed to do otherwise.
+
+# General Guidelines for Coding
+
+When building something from scratch, you should:
+
+- Understand the user's requirements.
+- Ask the user for clarification if there is anything unclear.
+- Design the architecture and make a plan for the implementation.
+- Write the code in a modular and maintainable way.
+
+Always use tools to implement your code changes:
+
+- Use `write`/`edit` to create or modify source files. Code that only appears in your text response is NOT saved to the file system and will not take effect.
+- Use `bash` to run and test your code after writing it.
+- Iterate: if tests fail, read the error, fix the code with `write`/`edit`, and re-test with `bash`.
+
+When working on an existing codebase, you should:
+
+- Understand the codebase by reading it with tools (`read`, `glob`, `grep`) before making changes. Identify the ultimate goal and the most important criteria to achieve the goal.
+- For a bug fix, you typically need to check error logs or failed tests, scan over the codebase to find the root cause, and figure out a fix. If user mentioned any failed tests, you should make sure they pass after the changes.
+- For a feature, you typically need to design the architecture, and write the code in a modular and maintainable way, with minimal intrusions to existing code. Add new tests if the project already has tests.
+- For a code refactoring, you typically need to update all the places that call the code you are refactoring if the interface changes. DO NOT change any existing logic especially in tests, focus only on fixing any errors caused by the interface changes.
+- Make MINIMAL changes to achieve the goal. This is very important to your performance.
+- Follow the coding style of existing code in the project.
+
+DO NOT run `git commit`, `git push`, `git reset`, `git rebase` and/or do any other git mutations unless explicitly asked to do so. Ask for confirmation each time when you need to do git mutations, even if the user has confirmed in earlier conversations.
+
+# General Guidelines for Research and Data Processing
+
+The user may ask you to research on certain topics, process or generate certain multimedia files. When doing such tasks, you must:
+
+- Understand the user's requirements thoroughly, ask for clarification before you start if needed.
+- Make plans before doing deep or wide research, to ensure you are always on track.
+- Search on the Internet if possible, with carefully-designed search queries to improve efficiency and accuracy.
+- Use proper tools or shell commands or Python packages to process or generate images, videos, PDFs, docs, spreadsheets, presentations, or other multimedia files. Detect if there are already such tools in the environment. If you have to install third-party tools/packages, you MUST ensure that they are installed in a virtual/isolated environment.
+- Once you generate or edit any images, videos or other media files, try to read it again before proceed, to ensure that the content is as expected.
+- Avoid installing or deleting anything to/from outside of the current working directory. If you have to do so, ask the user for confirmation.
+
+# Working Environment
+
+## Operating System
+
+The operating environment is not in a sandbox. Any actions you do will immediately affect the user's system. So you MUST be extremely cautious. Unless being explicitly instructed to do so, you should never access (read/write/execute) files outside of the working directory.
+
+## Working Directory
+
+The working directory should be considered as the project root if you are instructed to perform tasks on the project. Every file system operation will be relative to the working directory if you do not explicitly specify the absolute path. Tools may require absolute paths for some parameters, IF SO, YOU MUST use absolute paths for these parameters.
+
+# Project Information
+
+Markdown files named `AGENTS.md` usually contain the background, structure, coding styles, user preferences and other relevant information about the project. You should use this information to understand the project and the user's preferences. `AGENTS.md` files may exist at different locations in the project, but typically there is one in the project root.
+
+> Why `AGENTS.md`?
+>
+> `README.md` files are for humans: quick starts, project descriptions, and contribution guidelines. `AGENTS.md` complements this by containing the extra, sometimes detailed context coding agents need: build steps, tests, and conventions that might clutter a README or aren’t relevant to human contributors.
+>
+> We intentionally kept it separate to:
+>
+> - Give agents a clear, predictable place for instructions.
+> - Keep `README`s concise and focused on human contributors.
+> - Provide precise, agent-focused guidance that complements existing `README` and docs.
+If the `AGENTS.md` is empty or insufficient, you may check `README`/`README.md` files or `AGENTS.md` files in subdirectories for more information about specific parts of the project.
+
+If you modified any files/styles/structures/configurations/workflows/... mentioned in `AGENTS.md` files, you MUST update the corresponding `AGENTS.md` files to keep them up-to-date.
+
+# Ultimate Reminders
+
+At any time, you should be HELPFUL, CONCISE, and ACCURATE. Be thorough in your actions — test what you build, verify what you change — not in your explanations.
+
+- Never diverge from the requirements and the goals of the task you work on. Stay on track.
+- Never give the user more than what they want.
+- Try your best to avoid any hallucination. Do fact checking before providing any factual information.
+- Think about the best approach, then take action decisively.
+- Do not give up too early.
+- ALWAYS, keep it stupidly simple. Do not overcomplicate things.
+- When the task requires creating or modifying files, always use tools to do so. Never treat displaying code in your response as a substitute for actually writing it to the file system.

packages/core/src/session/runner/prompt/trinity.txt

@@ -0,0 +1,97 @@
+You are opencode, an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
+
+# Tone and style
+You should be concise, direct, and to the point. When you run a non-trivial bash command, you should explain what the command does and why you are running it, to make sure the user understands what you are doing (this is especially important when you are running a command that will make changes to the user's system).
+Remember that your output will be displayed on a command line interface. Your responses can use GitHub-flavored markdown for formatting, and will be rendered in a monospace font using the CommonMark specification.
+Output text to communicate with the user; all text you output outside of tool use is displayed to the user. Only use tools to complete tasks. Never use tools like Bash or code comments as means to communicate with the user during the session.
+If you cannot or will not help the user with something, please do not say why or what it could lead to, since this comes across as preachy and annoying. Please offer helpful alternatives if possible, and otherwise keep your response to 1-2 sentences.
+Only use emojis if the user explicitly requests it. Avoid using emojis in all communication unless asked.
+IMPORTANT: You should minimize output tokens as much as possible while maintaining helpfulness, quality, and accuracy. Only address the specific query or task at hand, avoiding tangential information unless absolutely critical for completing the request. If you can answer in 1-3 sentences or a short paragraph, please do.
+IMPORTANT: You should NOT answer with unnecessary preamble or postamble (such as explaining your code or summarizing your action), unless the user asks you to.
+IMPORTANT: Keep your responses short, since they will be displayed on a command line interface. You MUST answer concisely with fewer than 4 lines (not including tool use or code generation), unless user asks for detail. Answer the user's question directly, without elaboration, explanation, or details. One word answers are best. Avoid introductions, conclusions, and explanations. You MUST avoid text before/after your response, such as "The answer is <answer>.", "Here is the content of the file..." or "Based on the information provided, the answer is..." or "Here is what I will do next...". Here are some examples to demonstrate appropriate verbosity:
+<example>
+user: 2 + 2
+assistant: 4
+</example>
+
+<example>
+user: what is 2+2?
+assistant: 4
+</example>
+
+<example>
+user: is 11 a prime number?
+assistant: Yes
+</example>
+
+<example>
+user: what command should I run to list files in the current directory?
+assistant: ls
+</example>
+
+<example>
+user: what command should I run to watch files in the current directory?
+assistant: [use the ls tool to list the files in the current directory, then read docs/commands in the relevant file to find out how to watch files]
+npm run dev
+</example>
+
+<example>
+user: How many golf balls fit inside a jetta?
+assistant: 150000
+</example>
+
+<example>
+user: what files are in the directory src/?
+assistant: [runs ls and sees foo.c, bar.c, baz.c]
+user: which file contains the implementation of foo?
+assistant: src/foo.c
+</example>
+
+<example>
+user: write tests for new feature
+assistant: [uses grep or glob to find where similar tests are defined, then read relevant files one at a time (one tool per message, wait for each result), then edit or write to add tests]
+</example>
+
+# Proactiveness
+You are allowed to be proactive, but only when the user asks you to do something. You should strive to strike a balance between:
+1. Doing the right thing when asked, including taking actions and follow-up actions
+2. Not surprising the user with actions you take without asking
+For example, if the user asks you how to approach something, you should do your best to answer their question first, and not immediately jump into taking actions.
+3. Do not add additional code explanation summary unless requested by the user. After working on a file, just stop, rather than providing an explanation of what you did.
+
+# Following conventions
+When making changes to files, first understand the file's code conventions. Mimic code style, use existing libraries and utilities, and follow existing patterns.
+- NEVER assume that a given library is available, even if it is well known. Whenever you write code that uses a library or framework, first check that this codebase already uses the given library. For example, you might look at neighboring files, or check the package.json (or cargo.toml, and so on depending on the language).
+- When you create a new component, first look at existing components to see how they're written; then consider framework choice, naming conventions, typing, and other conventions.
+- When you edit a piece of code, first look at the code's surrounding context (especially its imports) to understand the code's choice of frameworks and libraries. Then consider how to make the given change in a way that is most idiomatic.
+- Always follow security best practices. Never introduce code that exposes or logs secrets and keys. Never commit secrets or keys to the repository.
+
+# Code style
+- IMPORTANT: DO NOT ADD ***ANY*** COMMENTS unless asked
+
+# Doing tasks
+The user will primarily request you perform software engineering tasks. This includes solving bugs, adding new functionality, refactoring code, explaining code, and more. For these tasks the following steps are recommended:
+- Use the available search tools to understand the codebase and the user's query. Use one tool per message; after each result, decide the next step and call one tool again.
+- Implement the solution using all tools available to you
+- Verify the solution if possible with tests. NEVER assume specific test framework or test script. Check the README or search codebase to determine the testing approach.
+- VERY IMPORTANT: When you have completed a task, you MUST run the lint and typecheck commands (e.g. npm run lint, npm run typecheck, ruff, etc.) with Bash if they were provided to you to ensure your code is correct. If you are unable to find the correct command, ask the user for the command to run and if they supply it, proactively suggest writing it to AGENTS.md so that you will know to run it next time.
+NEVER commit changes unless the user explicitly asks you to. It is VERY IMPORTANT to only commit when explicitly asked, otherwise the user will feel that you are being too proactive.
+
+- Tool results and user messages may include <system-reminder> tags. <system-reminder> tags contain useful information and reminders. They are NOT part of the user's provided input or the tool result.
+
+# Tool usage policy
+- When doing file search, prefer to use the Task tool in order to reduce context usage.
+- Use exactly one tool per assistant message. After each tool call, wait for the result before continuing.
+- When the user's request is vague, use the question tool to clarify before reading files or making changes.
+- Avoid repeating the same tool with the same parameters once you have useful results. Use the result to take the next step (e.g. pick one match, read that file, then act); do not search again in a loop.
+
+You MUST answer concisely with fewer than 4 lines of text (not including tool use or code generation), unless user asks for detail.
+
+# Code References
+
+When referencing specific functions or pieces of code include the pattern `file_path:line_number` to allow the user to easily navigate to the source code location.
+
+<example>
+user: Where are errors from the client handled?
+assistant: Clients are marked as failed in the `connectToServer` function in src/services/process.ts:712.
+</example>

packages/core/src/session/runner/system-prompt.ts

@@ -0,0 +1,30 @@
+export * as SessionRunnerSystemPrompt from "./system-prompt"
+
+import type { Model } from "@opencode-ai/llm"
+
+import PROMPT_ANTHROPIC from "./prompt/anthropic.txt"
+import PROMPT_BEAST from "./prompt/beast.txt"
+import PROMPT_CODEX from "./prompt/codex.txt"
+import PROMPT_DEFAULT from "./prompt/default.txt"
+import PROMPT_GEMINI from "./prompt/gemini.txt"
+import PROMPT_GPT from "./prompt/gpt.txt"
+import PROMPT_KIMI from "./prompt/kimi.txt"
+import PROMPT_TRINITY from "./prompt/trinity.txt"
+
+export function provider(model: Model) {
+  const id = model.id.toLowerCase()
+  if (id.includes("gpt-4") || id.includes("o1") || id.includes("o3")) return normalize(PROMPT_BEAST)
+  if (id.includes("gpt")) {
+    if (id.includes("codex")) return normalize(PROMPT_CODEX)
+    return normalize(PROMPT_GPT)
+  }
+  if (id.includes("gemini-")) return normalize(PROMPT_GEMINI)
+  if (id.includes("claude")) return normalize(PROMPT_ANTHROPIC)
+  if (id.includes("trinity")) return normalize(PROMPT_TRINITY)
+  if (id.includes("kimi")) return normalize(PROMPT_KIMI)
+  return normalize(PROMPT_DEFAULT)
+}
+
+function normalize(prompt: string) {
+  return prompt.replaceAll("\r\n", "\n")
+}

packages/core/src/session/runtime-coordinator.ts

@@ -1,47 +0,0 @@
-export * as SessionRuntimeCoordinator from "./runtime-coordinator"
-
-import { Context, Effect, Layer } from "effect"
-import { makeGlobalNode } from "../effect/app-node"
-import { SessionRunner } from "./runner"
-import { SessionRunCoordinator } from "./run-coordinator"
-import { SessionSchema } from "./schema"
-
-type Drain = (force: boolean) => Effect.Effect<void, SessionRunner.RunError>
-
-export interface Interface {
-  readonly active: Effect.Effect<ReadonlySet<SessionSchema.ID>>
-  readonly run: (sessionID: SessionSchema.ID, drain: Drain) => Effect.Effect<void, SessionRunner.RunError>
-  readonly wake: (sessionID: SessionSchema.ID, drain: Drain) => Effect.Effect<void>
-  readonly interrupt: (sessionID: SessionSchema.ID) => Effect.Effect<void>
-  readonly wait: (sessionID: SessionSchema.ID) => Effect.Effect<void>
-}
-
-export class Service extends Context.Service<Service, Interface>()("@opencode/v2/SessionRuntimeCoordinator") {}
-
-export const layer = Layer.effect(
-  Service,
-  Effect.gen(function* () {
-    const drains = new Map<SessionSchema.ID, Drain>()
-    const coordinator = yield* SessionRunCoordinator.make<SessionSchema.ID, SessionRunner.RunError>({
-      drain: Effect.fnUntraced(function* (sessionID, force) {
-        const drain = drains.get(sessionID)
-        if (!drain) return yield* Effect.die(`No SessionRuntime drain registered for ${sessionID}`)
-        return yield* drain(force)
-      }),
-    })
-
-    return Service.of({
-      active: coordinator.active,
-      run: (sessionID, drain) =>
-        Effect.sync(() => drains.set(sessionID, drain)).pipe(Effect.andThen(coordinator.run(sessionID))),
-      wake: (sessionID, drain) =>
-        Effect.sync(() => drains.set(sessionID, drain)).pipe(Effect.andThen(coordinator.wake(sessionID))),
-      interrupt: coordinator.interrupt,
-      wait: coordinator.awaitIdle,
-    })
-  }),
-)
-
-export const defaultLayer = layer
-
-export const node = makeGlobalNode({ service: Service, layer, deps: [] })

packages/core/src/session/runtime.ts

@@ -1,165 +0,0 @@
-export * as SessionRuntime from "./runtime"
-
-import { Context, Effect, Layer } from "effect"
-import { Database } from "../database/database"
-import { EventV2 } from "../event"
-import { Location } from "../location"
-import { PromptInput } from "@opencode-ai/schema/prompt-input"
-import { SessionMessage } from "./message"
-import { Prompt } from "./prompt"
-import { SessionInput } from "./input"
-import { SessionRevert } from "./revert"
-import { SessionRunner } from "./runner"
-import * as SessionRunnerLLM from "./runner/llm"
-import { SessionRuntimeCoordinator } from "./runtime-coordinator"
-import { SessionSchema } from "./schema"
-import { Snapshot } from "../snapshot"
-import { FSUtil } from "../fs-util"
-import { makeLocationNode } from "../effect/app-node"
-import { SessionV2 } from "../session"
-import {
-  BusyError,
-  MessageNotFoundError,
-  NotFoundError,
-  PromptConflictError,
-  type RevertState,
-} from "../session"
-
-export interface Interface {
-  readonly prompt: (input: {
-    id?: SessionMessage.ID
-    sessionID: SessionSchema.ID
-    prompt: PromptInput.Prompt
-    delivery?: SessionInput.Delivery
-    resume?: boolean
-  }) => Effect.Effect<SessionInput.Admitted, NotFoundError | PromptConflictError>
-  readonly wait: (id: SessionSchema.ID) => Effect.Effect<void, NotFoundError>
-  readonly active: Effect.Effect<ReadonlySet<SessionSchema.ID>>
-  readonly resume: (sessionID: SessionSchema.ID) => Effect.Effect<void, NotFoundError | SessionRunner.RunError>
-  readonly interrupt: (sessionID: SessionSchema.ID) => Effect.Effect<void>
-  readonly revert: {
-    readonly stage: (input: {
-      sessionID: SessionSchema.ID
-      messageID: SessionMessage.ID
-      files?: boolean
-    }) => Effect.Effect<RevertState, NotFoundError | MessageNotFoundError | BusyError | Snapshot.Error>
-    readonly clear: (sessionID: SessionSchema.ID) => Effect.Effect<void, NotFoundError | BusyError | Snapshot.Error>
-    readonly commit: (sessionID: SessionSchema.ID) => Effect.Effect<void, NotFoundError | BusyError>
-  }
-}
-
-export class Service extends Context.Service<Service, Interface>()("@opencode/v2/SessionRuntime") {}
-
-export const layer = Layer.effect(
-  Service,
-  Effect.gen(function* () {
-    const database = yield* Database.Service
-    const db = database.db
-    const events = yield* EventV2.Service
-    const location = yield* Location.Service
-    const sessions = yield* SessionV2.Service
-    const runner = yield* SessionRunner.Service
-    const snapshot = yield* Snapshot.Service
-    const coordinator = yield* SessionRuntimeCoordinator.Service
-
-    const local = Effect.fn("SessionRuntime.local")(function* (sessionID: SessionSchema.ID) {
-      const session = yield* sessions.get(sessionID)
-      if (session.location.directory !== location.directory || session.location.workspaceID !== location.workspaceID)
-        return yield* new NotFoundError({ sessionID })
-      return session
-    })
-
-    const drain = (sessionID: SessionSchema.ID) =>
-      Effect.fnUntraced(function* (force: boolean) {
-        yield* local(sessionID).pipe(Effect.orDie)
-        return yield* runner.run({ sessionID, force })
-      })
-
-    return Service.of({
-      prompt: Effect.fn("SessionRuntime.prompt")((input) =>
-        Effect.uninterruptible(
-          Effect.gen(function* () {
-            yield* local(input.sessionID)
-            const prompt = resolvePrompt(input.prompt)
-            const messageID = input.id ?? SessionMessage.ID.create()
-            const delivery = input.delivery ?? "steer"
-            const expected = { sessionID: input.sessionID, messageID, prompt, delivery }
-            const admitted = yield* SessionInput.admit(db, events, {
-              id: messageID,
-              sessionID: input.sessionID,
-              prompt,
-              delivery,
-            }).pipe(
-              Effect.catchDefect((defect) =>
-                defect instanceof SessionInput.LifecycleConflict
-                  ? new PromptConflictError({ sessionID: input.sessionID, messageID })
-                  : Effect.die(defect),
-              ),
-            )
-            if (!SessionInput.equivalent(admitted, expected))
-              return yield* new PromptConflictError({ sessionID: input.sessionID, messageID })
-            if (input.resume !== false) yield* coordinator.wake(admitted.sessionID, drain(admitted.sessionID))
-            return admitted
-          }),
-        ),
-      ),
-      wait: Effect.fn("SessionRuntime.wait")(function* (sessionID) {
-        yield* local(sessionID)
-        yield* coordinator.wait(sessionID)
-      }),
-      active: coordinator.active,
-      resume: Effect.fn("SessionRuntime.resume")(function* (sessionID) {
-        yield* local(sessionID)
-        yield* coordinator.run(sessionID, drain(sessionID))
-      }),
-      interrupt: Effect.fn("SessionRuntime.interrupt")(function* (sessionID) {
-        yield* local(sessionID)
-        yield* Effect.uninterruptible(coordinator.interrupt(sessionID))
-      }),
-      revert: {
-        stage: Effect.fn("SessionRuntime.revert.stage")(function* (input) {
-          const session = yield* local(input.sessionID)
-          if ((yield* coordinator.active).has(input.sessionID)) return yield* new BusyError({ sessionID: input.sessionID })
-          return yield* SessionRevert.stage({ session, messageID: input.messageID, files: input.files }).pipe(
-            Effect.provideService(Database.Service, database),
-            Effect.provideService(EventV2.Service, events),
-            Effect.provideService(Snapshot.Service, snapshot),
-          )
-        }),
-        clear: Effect.fn("SessionRuntime.revert.clear")(function* (sessionID) {
-          const session = yield* local(sessionID)
-          if ((yield* coordinator.active).has(sessionID)) return yield* new BusyError({ sessionID })
-          return yield* SessionRevert.clear(session).pipe(
-            Effect.provideService(EventV2.Service, events),
-            Effect.provideService(Snapshot.Service, snapshot),
-          )
-        }),
-        commit: Effect.fn("SessionRuntime.revert.commit")(function* (sessionID) {
-          const session = yield* local(sessionID)
-          if ((yield* coordinator.active).has(sessionID)) return yield* new BusyError({ sessionID })
-          return yield* SessionRevert.commit(session).pipe(Effect.provideService(EventV2.Service, events))
-        }),
-      },
-    })
-  }),
-)
-
-const resolvePrompt = (input: PromptInput.Prompt) =>
-  Prompt.make({
-    text: input.text,
-    agents: input.agents,
-    files: input.files?.map((file) => {
-      const dataMime = file.uri.match(/^data:([^;,]+)[;,]/i)?.[1]
-      const target = URL.canParse(file.uri) ? new URL(file.uri).pathname : (file.name ?? file.uri)
-      return {
-        ...file,
-        mime: dataMime ?? (target.endsWith("/") ? "application/x-directory" : FSUtil.mimeType(target)),
-      }
-    }),
-  })
-
-export const node = makeLocationNode({
-  service: Service,
-  layer,
-  deps: [Database.node, EventV2.node, Location.node, SessionV2.node, SessionRunnerLLM.node, Snapshot.node, SessionRuntimeCoordinator.node],
-})

packages/core/src/shell.ts

@@ -1,226 +1,300 @@
 export * as Shell from "./shell"
 
 import path from "path"
-import { spawn, type ChildProcess } from "child_process"
-import { readFile } from "fs/promises"
-import { statSync } from "fs"
-import { setTimeout as sleep } from "node:timers/promises"
-import { Flag } from "./flag/flag"
-import { FSUtil } from "./fs-util"
-import { which } from "./util/which"
+import { Context, Deferred, Duration, Effect, Fiber, Layer, Schema, Stream } from "effect"
+import { ChildProcess } from "effect/unstable/process"
+import { produce } from "immer"
+import { Shell } from "@opencode-ai/schema/shell"
+import { makeLocationNode } from "./effect/app-node"
+import { AppProcess } from "./process"
+import { Config } from "./config"
+import { EventV2 } from "./event"
+import { Location } from "./location"
+import { Global } from "./global"
+import { ShellSelect } from "./shell/select"
 
-const SIGKILL_TIMEOUT_MS = 200
-const META: Record<string, { deny?: boolean; login?: boolean; posix?: boolean; ps?: boolean }> = {
-  bash: { login: true, posix: true },
-  dash: { login: true, posix: true },
-  fish: { deny: true, login: true },
-  ksh: { login: true, posix: true },
-  nu: { deny: true },
-  powershell: { ps: true },
-  pwsh: { ps: true },
-  sh: { login: true, posix: true },
-  zsh: { login: true, posix: true },
+export class NotFoundError extends Schema.TaggedErrorClass<NotFoundError>()("Shell.NotFoundError", {
+  id: Shell.ID,
+}) {}
+
+// Exited processes stay observable (status, exit code, retained output) until removed explicitly.
+// Cap retention so abandoned commands do not accumulate unbounded state and output files.
+const EXITED_LIMIT = 25
+
+type Info = Shell.Info
+
+type Active = {
+  // Immutable snapshot; lifecycle updates replace it via immer `produce`.
+  info: Info
+  file: string
+  size: number
+  // Resolves with the terminal Info once the command exits, times out, or is killed. A wait
+  // started after termination resolves immediately from the already-completed deferred.
+  done: Deferred.Deferred<Info, NotFoundError>
+  timeoutFiber?: Fiber.Fiber<void, never>
 }
 
-export type Item = {
-  path: string
-  name: string
-  acceptable: boolean
+/**
+ * Location-owned non-interactive shell command process service.
+ *
+ * Each `create` spawns one shell command, captures combined stdout/stderr to a
+ * file, and returns an ID. Clients poll `get` for status and `output` for
+ * file-backed output by cursor. No session, message, or permission state lives
+ * here; callers (e.g. `ShellTool`) own that association and store the shell ID.
+ */
+export interface Interface {
+  readonly create: (input: Shell.CreateInput) => Effect.Effect<Shell.Info>
+  // Currently running commands only; exited shells are retained for get/output but excluded here.
+  readonly list: () => Effect.Effect<Shell.Info[]>
+  readonly get: (id: Shell.ID) => Effect.Effect<Shell.Info, NotFoundError>
+  // Resolves once the command reaches a terminal status, returning its final Info. Fails with
+  // NotFoundError if the command is unknown or is removed before it terminates.
+  readonly wait: (id: Shell.ID) => Effect.Effect<Shell.Info, NotFoundError>
+  readonly output: (id: Shell.ID, input?: Shell.OutputInput) => Effect.Effect<Shell.Output, NotFoundError>
+  readonly remove: (id: Shell.ID) => Effect.Effect<void, NotFoundError>
 }
 
-export async function killTree(proc: ChildProcess, opts?: { exited?: () => boolean }): Promise<void> {
-  const pid = proc.pid
-  if (!pid || opts?.exited?.()) return
+export class Service extends Context.Service<Service, Interface>()("@opencode/v2/Shell") {}
 
-  if (process.platform === "win32") {
-    await new Promise<void>((resolve) => {
-      const killer = spawn("taskkill", ["/pid", String(pid), "/f", "/t"], {
-        stdio: "ignore",
-        windowsHide: true,
-      })
-      killer.once("exit", () => resolve())
-      killer.once("error", () => resolve())
+export const layer = Layer.effect(
+  Service,
+  Effect.gen(function* () {
+    const events = yield* EventV2.Service
+    const location = yield* Location.Service
+    const config = yield* Config.Service
+    const global = yield* Global.Service
+    const appProcess = yield* AppProcess.Service
+    const context = yield* Effect.context()
+    const runFork = Effect.runForkWith(context)
+    const sessions = new Map<string, Active>()
+    const exitOrder: string[] = []
+
+    const outputDir = path.join(global.data, "shell", location.project.id)
+    const { mkdir, unlink } = yield* Effect.promise(() => import("fs/promises"))
+    const { createWriteStream, createReadStream } = yield* Effect.promise(() => import("fs"))
+    yield* Effect.promise(() => mkdir(outputDir, { recursive: true }))
+
+    yield* Effect.addFinalizer(() =>
+      Effect.gen(function* () {
+        for (const session of sessions.values()) {
+          if (session.timeoutFiber) yield* Fiber.interrupt(session.timeoutFiber)
+          // Unblock waiters still pending at teardown; succeed is a no-op once already resolved.
+          yield* Deferred.fail(session.done, new NotFoundError({ id: Shell.ID.make(session.info.id) }))
+        }
+        sessions.clear()
+        exitOrder.length = 0
+      }),
+    )
+
+    const require = Effect.fn("Shell.require")(function* (id: Shell.ID) {
+      const session = sessions.get(id)
+      if (!session) return yield* new NotFoundError({ id })
+      return session
     })
-    return
-  }
 
-  try {
-    process.kill(-pid, "SIGTERM")
-    await sleep(SIGKILL_TIMEOUT_MS)
-    if (!opts?.exited?.()) {
-      process.kill(-pid, "SIGKILL")
-    }
-  } catch {
-    proc.kill("SIGTERM")
-    await sleep(SIGKILL_TIMEOUT_MS)
-    if (!opts?.exited?.()) {
-      proc.kill("SIGKILL")
-    }
-  }
-}
+    const removeSession = Effect.fnUntraced(function* (id: Shell.ID) {
+      const session = sessions.get(id)
+      if (!session) return
+      sessions.delete(id)
+      const index = exitOrder.indexOf(id)
+      if (index !== -1) exitOrder.splice(index, 1)
+      if (session.timeoutFiber) yield* Fiber.interrupt(session.timeoutFiber)
+      // Unblock any wait still pending when the command is removed before it terminated.
+      yield* Deferred.fail(session.done, new NotFoundError({ id }))
+      yield* Effect.promise(() => unlink(session.file).catch(() => {}))
+      yield* events.publish(Shell.Event.Deleted, { id })
+    })
 
-function stat(file: string) {
-  return statSync(file, { throwIfNoEntry: false }) ?? undefined
-}
+    const remove = Effect.fn("Shell.remove")(function* (id: Shell.ID) {
+      yield* require(id)
+      yield* removeSession(id)
+    })
 
-function full(file: string) {
-  if (process.platform !== "win32") return file
-  const shell = FSUtil.windowsPath(file)
-  if (path.win32.dirname(shell) !== ".") {
-    if (shell.startsWith("/") && name(shell) === "bash") return gitbash() || shell
-    return shell
-  }
-  if (name(shell) === "bash") return gitbash() || which(shell) || shell
-  return which(shell) || shell
-}
+    const list = Effect.fn("Shell.list")(function* () {
+      return Array.from(sessions.values())
+        .filter((session) => session.info.status === "running")
+        .map((session) => session.info)
+    })
 
-function meta(file: string) {
-  return META[name(file)]
-}
+    const get = Effect.fn("Shell.get")(function* (id: Shell.ID) {
+      return (yield* require(id)).info
+    })
 
-function ok(file: string) {
-  return meta(file)?.deny !== true
-}
+    const wait = Effect.fn("Shell.wait")(function* (id: Shell.ID) {
+      return yield* Deferred.await((yield* require(id)).done)
+    })
 
-function rooted(file: string) {
-  return path.isAbsolute(FSUtil.windowsPath(file))
-}
+    const output = Effect.fn("Shell.output")(function* (id: Shell.ID, input?: Shell.OutputInput) {
+      const session = yield* require(id)
+      const cursor = input?.cursor ?? 0
+      const limit = input?.limit ?? 65536
+      if (cursor >= session.size) return { output: "", cursor: session.size, size: session.size, truncated: false }
+      const start = Math.max(0, cursor)
+      const length = Math.min(limit, session.size - start)
+      const buffer = Buffer.alloc(length)
+      const bytesRead = yield* Effect.promise(
+        () =>
+          new Promise<number>((resolve) => {
+            const stream = createReadStream(session.file, { start, end: start + length - 1 })
+            let offset = 0
+            stream.on("data", (chunk: string | Buffer) => {
+              const bytes = Buffer.from(chunk)
+              bytes.copy(buffer, offset)
+              offset += bytes.length
+            })
+            stream.on("end", () => resolve(offset))
+            stream.on("error", () => resolve(0))
+          }),
+      )
+      return {
+        output: buffer.subarray(0, bytesRead).toString("utf8"),
+        cursor: start + bytesRead,
+        size: session.size,
+        truncated: false,
+      }
+    })
 
-function resolve(file: string) {
-  const shell = full(file)
-  if (rooted(shell)) {
-    if (stat(shell)?.isFile()) return shell
-    return
-  }
-  return which(shell) ?? undefined
-}
+    const create = Effect.fn("Shell.create")(function* (input: Shell.CreateInput) {
+      const id = Shell.ID.ascending()
+      const cwd = input.cwd ?? location.directory
+      const configShell = Config.latest(yield* config.entries(), "shell")
+      const shell = ShellSelect.preferred(configShell)
+      const args = ShellSelect.args(shell, input.command, cwd)
+      const file = path.join(outputDir, `${id}.out`)
+      const env = {
+        ...process.env,
+        TERM: "xterm-256color",
+        OPENCODE_TERMINAL: "1",
+      } as Record<string, string>
 
-function win() {
-  return Array.from(
-    new Set(
-      [which("pwsh"), which("powershell"), gitbash(), process.env.COMSPEC || "cmd.exe"]
-        .filter((item): item is string => Boolean(item))
-        .map(full),
-    ),
-  )
-}
+      const info: Info = {
+        id,
+        status: "running",
+        command: input.command,
+        cwd,
+        shell,
+        file,
+        metadata: input.metadata ?? {},
+        time: { started: Date.now() },
+      }
 
-async function unix() {
-  const text = await readFile("/etc/shells", "utf8").catch(() => "")
-  if (text) return Array.from(new Set(text.split("\n").filter((line) => line.trim() && !line.startsWith("#"))))
-  return ["/bin/bash", "/bin/zsh", "/bin/sh"]
-}
+      // Spawn via AppProcess and stream combined output to the file. The handle is scope-bound, so
+      // the managing fiber keeps its scope open until the command terminates (it awaits `done` at the
+      // end). `create` returns once `ready` resolves with the registered session.
+      const ready = Deferred.makeUnsafe<Active, never>()
+      runFork(
+        Effect.scoped(
+          Effect.gen(function* () {
+            const handle = yield* appProcess.spawn(
+              ChildProcess.make(shell, args, {
+                cwd,
+                env,
+                stdin: "ignore",
+                detached: process.platform !== "win32",
+                forceKillAfter: Duration.seconds(3),
+              }),
+            )
+            const session: Active = {
+              info: produce(info, (draft) => {
+                draft.pid = handle.pid
+              }),
+              file,
+              size: 0,
+              done: Deferred.makeUnsafe<Info, NotFoundError>(),
+            }
+            sessions.set(id, session)
 
-function select(file: string | undefined, opts?: { acceptable?: boolean }) {
-  if (file && (!opts?.acceptable || ok(file))) {
-    const shell = resolve(file)
-    if (shell) return shell
-  }
-  if (process.platform === "win32") return win()[0]
-  return fallback()
-}
+            const stream = createWriteStream(file)
+            yield* Effect.promise(
+              () =>
+                new Promise<void>((resolve) => {
+                  stream.once("open", () => resolve())
+                  stream.once("error", () => resolve())
+                }),
+            )
 
-export function gitbash() {
-  if (process.platform !== "win32") return
-  if (Flag.OPENCODE_GIT_BASH_PATH) return Flag.OPENCODE_GIT_BASH_PATH
-  const git = which("git")
-  if (!git) return
-  const file = path.join(git, "..", "..", "bin", "bash.exe")
-  if (stat(file)?.size) return file
-}
+            const pump = handle.all.pipe(
+              Stream.runForEach((chunk: Uint8Array) =>
+                Effect.sync(() => {
+                  stream.write(chunk)
+                  session.size += chunk.length
+                }),
+              ),
+            )
+            runFork(pump.pipe(Effect.catch(() => Effect.void)))
 
-function fallback() {
-  if (process.platform === "darwin") return "/bin/zsh"
-  const bash = which("bash")
-  if (bash) return bash
-  return "/bin/sh"
-}
+            const finish = (status: Info["status"], exit?: number) =>
+              Effect.gen(function* () {
+                if (session.info.status !== "running") return
+                session.info = produce(session.info, (draft) => {
+                  draft.status = status
+                  if (exit !== undefined) draft.exit = exit
+                  draft.time.completed = Date.now()
+                })
+                stream.end()
+                // Resolve waiters with the terminal Info before any retention eviction, so an evicted
+                // session still reports success rather than the removal NotFoundError. This runs before
+                // the timeout-fiber interrupt below, which on the timeout path would otherwise cancel
+                // this very fiber (finish is invoked by the timeout fiber) before waiters are resolved.
+                yield* Deferred.succeed(session.done, session.info)
+                yield* events.publish(Shell.Event.Exited, {
+                  id,
+                  ...(exit !== undefined ? { exit } : {}),
+                  status,
+                })
+                exitOrder.push(id)
+                while (exitOrder.length > EXITED_LIMIT) {
+                  const oldest = exitOrder[0]
+                  if (!oldest) break
+                  yield* removeSession(Shell.ID.make(oldest))
+                }
+                // Cancel a pending timeout once the command exits on its own. Interrupting last avoids
+                // aborting finish when finish itself runs on the timeout fiber.
+                if (session.timeoutFiber) yield* Fiber.interrupt(session.timeoutFiber)
+              })
 
-export function name(file: string) {
-  if (process.platform === "win32") return path.win32.parse(FSUtil.windowsPath(file)).name.toLowerCase()
-  return path.basename(file).toLowerCase()
-}
+            if (input.timeout) {
+              session.timeoutFiber = runFork(
+                Effect.sleep(Duration.millis(input.timeout)).pipe(
+                  Effect.flatMap(() =>
+                    Effect.gen(function* () {
+                      yield* finish("timeout")
+                      yield* handle.kill().pipe(Effect.catch(() => Effect.void))
+                    }),
+                  ),
+                  Effect.catch(() => Effect.void),
+                ),
+              )
+            }
 
-export function login(file: string) {
-  return meta(file)?.login === true
-}
+            runFork(
+              handle.exitCode.pipe(
+                Effect.flatMap((code) => finish("exited", code)),
+                Effect.catch(() => Effect.void),
+              ),
+            )
 
-export function posix(file: string) {
-  return meta(file)?.posix === true
-}
+            yield* events.publish(Shell.Event.Created, { info })
+            yield* Deferred.succeed(ready, session)
+            // Hold the handle's scope open until the command terminates; closing it earlier would
+            // release (kill) the process before its exit is observed.
+            yield* Deferred.await(session.done).pipe(Effect.catch(() => Effect.void))
+          }),
+        ).pipe(Effect.catch(() => Effect.void)),
+      )
 
-export function ps(file: string) {
-  return meta(file)?.ps === true
-}
+      const session = yield* Deferred.await(ready)
+      return session.info
+    })
 
-function info(file: string): Item {
-  const item = full(file)
-  const n = name(item)
-  return {
-    path: item,
-    name: resolve(n) ? n : item,
-    acceptable: ok(item),
-  }
-}
+    return Service.of({ create, list, get, wait, output, remove })
+  }),
+)
 
-export function args(file: string, command: string, cwd: string) {
-  const n = name(file)
-  if (n === "nu" || n === "fish") return ["-c", command]
-  if (n === "zsh") {
-    return [
-      "-l",
-      "-c",
-      `
-        [[ -f ~/.zshenv ]] && source ~/.zshenv >/dev/null 2>&1 || true
-        [[ -f "\${ZDOTDIR:-$HOME}/.zshrc" ]] && source "\${ZDOTDIR:-$HOME}/.zshrc" >/dev/null 2>&1 || true
-        cd -- "$1"
-        eval ${JSON.stringify(command)}
-      `,
-      "opencode",
-      cwd,
-    ]
-  }
-  if (n === "bash") {
-    return [
-      "-l",
-      "-c",
-      `
-        shopt -s expand_aliases
-        [[ -f ~/.bashrc ]] && source ~/.bashrc >/dev/null 2>&1 || true
-        cd -- "$1"
-        eval ${JSON.stringify(command)}
-      `,
-      "opencode",
-      cwd,
-    ]
-  }
-  if (n === "cmd") return ["/c", command]
-  if (ps(file)) return ["-NoProfile", "-Command", command]
-  return ["-c", command]
-}
+export const locationLayer = layer.pipe(Layer.provide(Config.locationLayer))
 
-let defaultPreferred: string | undefined
-let defaultAcceptable: string | undefined
-
-export function preferred(configShell?: string) {
-  if (configShell) return select(configShell)
-  defaultPreferred ??= select(process.env.SHELL)
-  return defaultPreferred
-}
-preferred.reset = () => {
-  defaultPreferred = undefined
-}
-
-export function acceptable(configShell?: string) {
-  if (configShell) return select(configShell, { acceptable: true })
-  defaultAcceptable ??= select(process.env.SHELL, { acceptable: true })
-  return defaultAcceptable
-}
-acceptable.reset = () => {
-  defaultAcceptable = undefined
-}
-
-export async function list(): Promise<Item[]> {
-  const shells = process.platform === "win32" ? win() : await unix()
-  return shells.filter((s) => resolve(s)).map(info)
-}
+export const node = makeLocationNode({
+  service: Service,
+  layer,
+  deps: [EventV2.node, Location.node, Config.node, Global.node, AppProcess.node],
+})

packages/core/src/shell/select.ts

@@ -0,0 +1,226 @@
+export * as ShellSelect from "./select"
+
+import path from "path"
+import { spawn, type ChildProcess } from "child_process"
+import { readFile } from "fs/promises"
+import { statSync } from "fs"
+import { setTimeout as sleep } from "node:timers/promises"
+import { Flag } from "../flag/flag"
+import { FSUtil } from "../fs-util"
+import { which } from "../util/which"
+
+const SIGKILL_TIMEOUT_MS = 200
+const META: Record<string, { deny?: boolean; login?: boolean; posix?: boolean; ps?: boolean }> = {
+  bash: { login: true, posix: true },
+  dash: { login: true, posix: true },
+  fish: { deny: true, login: true },
+  ksh: { login: true, posix: true },
+  nu: { deny: true },
+  powershell: { ps: true },
+  pwsh: { ps: true },
+  sh: { login: true, posix: true },
+  zsh: { login: true, posix: true },
+}
+
+export type Item = {
+  path: string
+  name: string
+  acceptable: boolean
+}
+
+export async function killTree(proc: ChildProcess, opts?: { exited?: () => boolean }): Promise<void> {
+  const pid = proc.pid
+  if (!pid || opts?.exited?.()) return
+
+  if (process.platform === "win32") {
+    await new Promise<void>((resolve) => {
+      const killer = spawn("taskkill", ["/pid", String(pid), "/f", "/t"], {
+        stdio: "ignore",
+        windowsHide: true,
+      })
+      killer.once("exit", () => resolve())
+      killer.once("error", () => resolve())
+    })
+    return
+  }
+
+  try {
+    process.kill(-pid, "SIGTERM")
+    await sleep(SIGKILL_TIMEOUT_MS)
+    if (!opts?.exited?.()) {
+      process.kill(-pid, "SIGKILL")
+    }
+  } catch {
+    proc.kill("SIGTERM")
+    await sleep(SIGKILL_TIMEOUT_MS)
+    if (!opts?.exited?.()) {
+      proc.kill("SIGKILL")
+    }
+  }
+}
+
+function stat(file: string) {
+  return statSync(file, { throwIfNoEntry: false }) ?? undefined
+}
+
+function full(file: string) {
+  if (process.platform !== "win32") return file
+  const shell = FSUtil.windowsPath(file)
+  if (path.win32.dirname(shell) !== ".") {
+    if (shell.startsWith("/") && name(shell) === "bash") return gitbash() || shell
+    return shell
+  }
+  if (name(shell) === "bash") return gitbash() || which(shell) || shell
+  return which(shell) || shell
+}
+
+function meta(file: string) {
+  return META[name(file)]
+}
+
+function ok(file: string) {
+  return meta(file)?.deny !== true
+}
+
+function rooted(file: string) {
+  return path.isAbsolute(FSUtil.windowsPath(file))
+}
+
+function resolve(file: string) {
+  const shell = full(file)
+  if (rooted(shell)) {
+    if (stat(shell)?.isFile()) return shell
+    return
+  }
+  return which(shell) ?? undefined
+}
+
+function win() {
+  return Array.from(
+    new Set(
+      [which("pwsh"), which("powershell"), gitbash(), process.env.COMSPEC || "cmd.exe"]
+        .filter((item): item is string => Boolean(item))
+        .map(full),
+    ),
+  )
+}
+
+async function unix() {
+  const text = await readFile("/etc/shells", "utf8").catch(() => "")
+  if (text) return Array.from(new Set(text.split("\n").filter((line) => line.trim() && !line.startsWith("#"))))
+  return ["/bin/bash", "/bin/zsh", "/bin/sh"]
+}
+
+function select(file: string | undefined, opts?: { acceptable?: boolean }) {
+  if (file && (!opts?.acceptable || ok(file))) {
+    const shell = resolve(file)
+    if (shell) return shell
+  }
+  if (process.platform === "win32") return win()[0]
+  return fallback()
+}
+
+export function gitbash() {
+  if (process.platform !== "win32") return
+  if (Flag.OPENCODE_GIT_BASH_PATH) return Flag.OPENCODE_GIT_BASH_PATH
+  const git = which("git")
+  if (!git) return
+  const file = path.join(git, "..", "..", "bin", "bash.exe")
+  if (stat(file)?.size) return file
+}
+
+function fallback() {
+  if (process.platform === "darwin") return "/bin/zsh"
+  const bash = which("bash")
+  if (bash) return bash
+  return "/bin/sh"
+}
+
+export function name(file: string) {
+  if (process.platform === "win32") return path.win32.parse(FSUtil.windowsPath(file)).name.toLowerCase()
+  return path.basename(file).toLowerCase()
+}
+
+export function login(file: string) {
+  return meta(file)?.login === true
+}
+
+export function posix(file: string) {
+  return meta(file)?.posix === true
+}
+
+export function ps(file: string) {
+  return meta(file)?.ps === true
+}
+
+function info(file: string): Item {
+  const item = full(file)
+  const n = name(item)
+  return {
+    path: item,
+    name: resolve(n) ? n : item,
+    acceptable: ok(item),
+  }
+}
+
+export function args(file: string, command: string, cwd: string) {
+  const n = name(file)
+  if (n === "nu" || n === "fish") return ["-c", command]
+  if (n === "zsh") {
+    return [
+      "-l",
+      "-c",
+      `
+        [[ -f ~/.zshenv ]] && source ~/.zshenv >/dev/null 2>&1 || true
+        [[ -f "\${ZDOTDIR:-$HOME}/.zshrc" ]] && source "\${ZDOTDIR:-$HOME}/.zshrc" >/dev/null 2>&1 || true
+        cd -- "$1"
+        eval ${JSON.stringify(command)}
+      `,
+      "opencode",
+      cwd,
+    ]
+  }
+  if (n === "bash") {
+    return [
+      "-l",
+      "-c",
+      `
+        shopt -s expand_aliases
+        [[ -f ~/.bashrc ]] && source ~/.bashrc >/dev/null 2>&1 || true
+        cd -- "$1"
+        eval ${JSON.stringify(command)}
+      `,
+      "opencode",
+      cwd,
+    ]
+  }
+  if (n === "cmd") return ["/c", command]
+  if (ps(file)) return ["-NoProfile", "-Command", command]
+  return ["-c", command]
+}
+
+let defaultPreferred: string | undefined
+let defaultAcceptable: string | undefined
+
+export function preferred(configShell?: string) {
+  if (configShell) return select(configShell)
+  defaultPreferred ??= select(process.env.SHELL)
+  return defaultPreferred
+}
+preferred.reset = () => {
+  defaultPreferred = undefined
+}
+
+export function acceptable(configShell?: string) {
+  if (configShell) return select(configShell, { acceptable: true })
+  defaultAcceptable ??= select(process.env.SHELL, { acceptable: true })
+  return defaultAcceptable
+}
+acceptable.reset = () => {
+  defaultAcceptable = undefined
+}
+
+export async function list(): Promise<Item[]> {
+  const shells = process.platform === "win32" ? win() : await unix()
+  return shells.filter((s) => resolve(s)).map(info)
+}

packages/core/src/tool/builtins.ts

@@ -2,7 +2,7 @@ export * as BuiltInTools from "./builtins"
 
 import { makeLocationNode } from "../effect/app-node"
 import { Layer } from "effect"
-import { BashTool } from "./bash"
+import { ShellTool } from "./shell"
 import { ApplyPatchTool } from "./apply-patch"
 import { EditTool } from "./edit"
 import { GlobTool } from "./glob"
@@ -16,8 +16,7 @@ import { WebFetchTool } from "./webfetch"
 import { WebSearchTool } from "./websearch"
 import { WriteTool } from "./write"
 import { FSUtil } from "../fs-util"
-import { AppProcess } from "../process"
-import { Config } from "../config"
+import { Shell } from "../shell"
 import { Location } from "../location"
 import { LocationMutation } from "../location-mutation"
 import { FileMutation } from "../file-mutation"
@@ -45,7 +44,7 @@ import { httpClient } from "../effect/app-node-platform"
  */
 export const locationLayer = Layer.mergeAll(
   ApplyPatchTool.layer,
-  BashTool.layer,
+  ShellTool.layer,
   EditTool.layer,
   GlobTool.layer,
   GrepTool.layer,
@@ -64,8 +63,7 @@ export const node = makeLocationNode({
   deps: [
     ToolRegistry.toolsNode,
     FSUtil.node,
-    AppProcess.node,
-    Config.node,
+    Shell.node,
     Location.node,
     LocationMutation.node,
     FileMutation.node,

packages/core/src/tool/shell.ts

@@ -1,19 +1,17 @@
-export * as BashTool from "./bash"
+export * as ShellTool from "./shell"
 
 import path from "path"
 import { ToolFailure } from "@opencode-ai/llm"
-import { Duration, Effect, Layer, Schema } from "effect"
-import { ChildProcess } from "effect/unstable/process"
-import { Config } from "../config"
+import { Effect, Layer, Schema } from "effect"
 import { FSUtil } from "../fs-util"
 import { LocationMutation } from "../location-mutation"
-import { AppProcess } from "../process"
 import { PermissionV2 } from "../permission"
 import { PositiveInt } from "../schema"
+import { Shell } from "../shell"
 import { Tool } from "./tool"
 import { Tools } from "./tools"
 
-export const name = "bash"
+export const name = "shell"
 export const DEFAULT_TIMEOUT_MS = 2 * 60 * 1_000
 export const MAX_TIMEOUT_MS = 10 * 60 * 1_000
 export const MAX_CAPTURE_BYTES = 1024 * 1024
@@ -44,8 +42,6 @@ const Output = Schema.Struct({
 
 type Output = typeof Output.Type
 
-const defaultShell = () => (process.platform === "win32" ? (process.env.COMSPEC ?? "cmd.exe") : "/bin/sh")
-
 const modelOutput = (output: Output) => {
   const warnings = output.warnings?.length
     ? `\n\nWarnings:\n${output.warnings.map((warning) => `- ${warning}`).join("\n")}`
@@ -54,9 +50,6 @@ const modelOutput = (output: Output) => {
   return `${warnings.trimStart()}${warnings ? "\n\n" : ""}Command exited with code ${output.exit}.`
 }
 
-const isTimeout = (error: AppProcess.AppProcessError) =>
-  error.cause instanceof Error && error.cause.message === "Timed out"
-
 /**
  * Minimal V2 core shell boundary. Keep parity debt visible without pulling the
  * legacy shell runtime into core.
@@ -93,8 +86,7 @@ export const layer = Layer.effectDiscard(
     const tools = yield* Tools.Service
     const mutation = yield* LocationMutation.Service
     const fs = yield* FSUtil.Service
-    const appProcess = yield* AppProcess.Service
-    const config = yield* Config.Service
+    const shell = yield* Shell.Service
     const permission = yield* PermissionV2.Service
 
     yield* tools
@@ -131,7 +123,7 @@ export const layer = Layer.effectDiscard(
                 })
               const warnings = externalCommandDirectories(input.command, target.canonical).map(
                 (directory) =>
-                  `Command argument references external directory ${path.join(directory, "*").replaceAll("\\", "/")}. Bash runs with host-user filesystem, process, and network authority; this scan is advisory only.`,
+                  `Command argument references external directory ${path.join(directory, "*").replaceAll("\\", "/")}. Shell runs with host-user filesystem, process, and network authority; this scan is advisory only.`,
               )
               yield* permission.assert({
                 action: name,
@@ -145,30 +137,20 @@ export const layer = Layer.effectDiscard(
               if ((yield* fs.stat(target.canonical)).type !== "Directory")
                 return yield* Effect.fail(new Error(`Working directory is not a directory: ${target.canonical}`))
 
-              const entries = yield* config.entries()
-              const shell =
-                Object.assign({}, ...entries.flatMap((entry) => (entry.type === "document" ? [entry.info] : [])))
-                  .shell ?? defaultShell()
-              const command = ChildProcess.make(input.command, [], {
-                cwd: target.canonical,
-                shell,
-                stdin: "ignore",
-                detached: process.platform !== "win32",
-                forceKillAfter: Duration.seconds(3),
-              })
+              // Delegate spawning, combined-output capture, timeout, and exit tracking to the Shell
+              // service. The full output is captured to a file; we read a bounded page for the model
+              // and point the agent at the file when it overflows the model cap.
               const timeout = input.timeout ?? DEFAULT_TIMEOUT_MS
-              const result = yield* appProcess
-                .run(command, {
-                  combineOutput: true,
-                  timeout: Duration.millis(timeout),
-                  maxOutputBytes: MAX_CAPTURE_BYTES,
-                })
-                .pipe(
-                  Effect.catchTag("AppProcessError", (error) =>
-                    isTimeout(error) ? Effect.succeed(undefined) : Effect.fail(error),
-                  ),
-                )
-              if (!result) {
+              const info = yield* shell.create({
+                command: input.command,
+                cwd: target.canonical,
+                timeout,
+                metadata: { sessionID: context.sessionID },
+              })
+              const final = yield* shell.wait(info.id)
+              const page = yield* shell.output(info.id, { limit: MAX_CAPTURE_BYTES })
+
+              if (final.status === "timeout") {
                 return {
                   output: `Command exceeded timeout of ${timeout} ms. Retry with a larger timeout if the command is expected to take longer.`,
                   truncated: false,
@@ -177,14 +159,13 @@ export const layer = Layer.effectDiscard(
                 }
               }
 
-              const output = result.output?.toString("utf8") || "(no output)"
-              const notice = result.outputTruncated
-                ? "[output capture truncated at the in-memory safety limit]"
-                : undefined
+              const truncated = page.size > page.cursor
+              const body = page.output || "(no output)"
+              const notice = truncated ? `\n\n[output truncated; full output saved to: ${final.file}]` : ""
               return {
-                exit: result.exitCode,
-                output: notice ? `${output}\n\n${notice}` : output,
-                truncated: result.outputTruncated === true,
+                exit: final.exit,
+                output: `${body}${notice}`,
+                truncated,
                 ...(warnings.length ? { warnings } : {}),
               }
             }).pipe(Effect.mapError(() => new ToolFailure({ message: `Unable to execute command: ${input.command}` }))),

packages/core/src/tool/subagent.ts

@@ -7,7 +7,6 @@ import { BackgroundJob } from "../background-job"
 import { EventV2 } from "../event"
 import { LocationServiceMap } from "../location-service-map"
 import { SessionV2 } from "../session"
-import { SessionRuntime } from "../session/runtime"
 import { SessionEvent } from "../session/event"
 import { SessionMessage } from "../session/message"
 import { SessionSchema } from "../session/schema"
@@ -117,9 +116,7 @@ export const layer = Layer.effectDiscard(
                 .pipe(
                   Effect.mapError(() => new ToolFailure({ message: `Parent session not found: ${context.sessionID}` })),
                 )
-              const locationLayer = locations.get(parent.location)
-              const agents = yield* AgentV2.Service.pipe(Effect.provide(locationLayer))
-              const runtime = yield* SessionRuntime.Service.pipe(Effect.provide(locationLayer))
+              const agents = yield* AgentV2.Service.pipe(Effect.provide(locations.get(parent.location)))
               const agent = yield* agents.resolve(input.agent)
               if (agent === undefined) return yield* new ToolFailure({ message: `Unknown agent: ${input.agent}` })
               if (agent.mode === "primary")
@@ -144,8 +141,8 @@ export const layer = Layer.effectDiscard(
 
               const run = Effect.gen(function* () {
                 // The child session owns its agent/model (set at create); prompt only admits input.
-                yield* runtime.prompt({ sessionID: child.id, prompt: { text: input.prompt }, resume: false })
-                yield* runtime.resume(child.id)
+                yield* sessions.prompt({ sessionID: child.id, prompt: { text: input.prompt }, resume: false })
+                yield* sessions.resume(child.id)
                 return yield* latestAssistantText(child.id)
               })
 
@@ -169,7 +166,7 @@ export const layer = Layer.effectDiscard(
                 jobs.waitForPromotion(child.id),
               ).pipe(
                 Effect.onInterrupt(() =>
-                  Effect.all([runtime.interrupt(child.id), jobs.cancel(child.id)], { discard: true }),
+                  Effect.all([sessions.interrupt(child.id), jobs.cancel(child.id)], { discard: true }),
                 ),
               )
               if (result?.metadata?.background === true)

packages/core/src/v1/config/migrate.ts

@@ -79,8 +79,9 @@ function permissions(info?: ConfigPermissionV1.Info, tools?: Readonly<Record<str
     resource: "*",
     effect: enabled ? ("allow" as const) : ("deny" as const),
   }))
-  for (const [action, rule] of Object.entries(info ?? {})) {
+  for (const [key, rule] of Object.entries(info ?? {})) {
     if (!rule) continue
+    const action = normalizeAction(key)
     if (typeof rule === "string") {
       rules.push({ action, resource: "*", effect: rule })
       continue
@@ -90,8 +91,12 @@ function permissions(info?: ConfigPermissionV1.Info, tools?: Readonly<Record<str
   return rules.length ? rules : undefined
 }
 
+// Map v1 permission/tool keys onto their renamed v2 tool actions so migrated rules keep matching.
 function normalizeAction(action: string) {
-  return action === "write" || action === "patch" ? "edit" : action
+  if (action === "write" || action === "patch") return "edit"
+  if (action === "task") return "subagent"
+  if (action === "bash") return "shell"
+  return action
 }
 
 function agents(info: typeof ConfigV1.Info.Type) {

packages/core/test/agent.test.ts

@@ -122,6 +122,7 @@ describe("AgentV2", () => {
         "summary",
         "title",
       ])
+      expect((yield* agent.get(AgentV2.defaultID))?.system).toBeUndefined()
       for (const item of agents) {
         expect(item.permissions.some((rule) => rule.action === "bash" && rule.effect !== "deny")).toBe(false)
       }

packages/core/test/config/config.test.ts

@@ -143,6 +143,27 @@ describe("Config", () => {
     }),
   )
 
+  it.effect("normalizes renamed permission actions when migrating v1 permissions", () =>
+    Effect.sync(() => {
+      expect(
+        ConfigMigrateV1.migrate({
+          permission: {
+            task: "ask",
+            bash: { "git status": "allow", "*": "deny" },
+            write: "deny",
+            read: "allow",
+          },
+        }).permissions,
+      ).toEqual([
+        { action: "subagent", resource: "*", effect: "ask" },
+        { action: "shell", resource: "git status", effect: "allow" },
+        { action: "shell", resource: "*", effect: "deny" },
+        { action: "edit", resource: "*", effect: "deny" },
+        { action: "read", resource: "*", effect: "allow" },
+      ])
+    }),
+  )
+
   it.live("returns an empty configuration when directory files do not exist", () =>
     Effect.acquireRelease(
       Effect.promise(() => tmpdir()),
@@ -566,7 +587,7 @@ describe("Config", () => {
             expect(documents[0]?.info.snapshots).toBe(false)
             expect(documents[0]?.info.share).toBe("auto")
             expect(documents[0]?.info.permissions).toEqual([
-              { action: "bash", resource: "*", effect: "ask" },
+              { action: "shell", resource: "*", effect: "ask" },
               { action: "edit", resource: "*.md", effect: "allow" },
               { action: "edit", resource: "*", effect: "deny" },
               { action: "question", resource: "*", effect: "deny" },

packages/core/test/session-compact.test.ts

@@ -0,0 +1,107 @@
+import { describe, expect } from "bun:test"
+import { LLMClient, LLMEvent, Model, type LLMRequest } from "@opencode-ai/llm"
+import { OpenAIChat } from "@opencode-ai/llm/protocols"
+import { Config } from "@opencode-ai/core/config"
+import { Database } from "@opencode-ai/core/database/database"
+import { EventV2 } from "@opencode-ai/core/event"
+import { Location } from "@opencode-ai/core/location"
+import { LocationServiceMap } from "@opencode-ai/core/location-service-map"
+import type { LocationServices } from "@opencode-ai/core/location-services"
+import { ProjectV2 } from "@opencode-ai/core/project"
+import { AbsolutePath } from "@opencode-ai/core/schema"
+import { SessionV2 } from "@opencode-ai/core/session"
+import { SessionCompaction } from "@opencode-ai/core/session/compaction"
+import { SessionEvent } from "@opencode-ai/core/session/event"
+import { SessionMessage } from "@opencode-ai/core/session/message"
+import { Prompt } from "@opencode-ai/core/session/prompt"
+import { SessionProjector } from "@opencode-ai/core/session/projector"
+import { SessionExecution } from "@opencode-ai/core/session/execution"
+import { SessionRunnerModel } from "@opencode-ai/core/session/runner/model"
+import { SessionStore } from "@opencode-ai/core/session/store"
+import { DateTime, Effect, Layer, LayerMap, Stream } from "effect"
+import { testEffect } from "./lib/effect"
+
+const location = Location.Ref.make({ directory: AbsolutePath.make("/project") })
+const model = Model.make({
+  id: "summary-model",
+  provider: "test",
+  route: OpenAIChat.route.with({ limits: { context: 10_000, output: 1_000 } }),
+})
+const projects = Layer.succeed(
+  ProjectV2.Service,
+  ProjectV2.Service.of({
+    resolve: (directory) => Effect.succeed({ id: ProjectV2.ID.global, directory }),
+    directories: () => Effect.succeed([]),
+    commit: () => Effect.void,
+  }),
+)
+let requests: LLMRequest[] = []
+const client = Layer.mock(LLMClient.Service)({
+  prepare: () => Effect.die("unused"),
+  stream: (request: LLMRequest) => {
+    requests.push(request)
+    return Stream.make(LLMEvent.textDelta({ id: "summary", text: "manual session summary" }))
+  },
+  generate: () => Effect.die("unused"),
+})
+const config = Layer.mock(Config.Service)({ entries: () => Effect.succeed([]) })
+const models = SessionRunnerModel.layerWith(() => Effect.succeed(model))
+const locations = Layer.effect(
+  LocationServiceMap.Service,
+  LayerMap.make(
+    () =>
+      // The test only needs the compaction location service used by SessionV2.compact.
+      // oxlint-disable-next-line typescript-eslint/no-unsafe-type-assertion
+      SessionCompaction.layer.pipe(
+        Layer.provide(client),
+        Layer.provide(config),
+        Layer.provide(models),
+      ) as unknown as Layer.Layer<LocationServices>,
+  ),
+)
+const sessions = SessionV2.layer.pipe(
+  Layer.provide(locations),
+  Layer.provide(EventV2.defaultLayer),
+  Layer.provide(Database.defaultLayer),
+  Layer.provide(SessionStore.defaultLayer),
+  Layer.provide(projects),
+  Layer.provide(SessionExecution.noopLayer),
+)
+const it = testEffect(
+  Layer.mergeAll(
+    Database.defaultLayer,
+    EventV2.defaultLayer,
+    projects,
+    SessionProjector.defaultLayer,
+    SessionStore.defaultLayer,
+    SessionExecution.noopLayer,
+    sessions,
+  ),
+)
+
+describe("SessionV2.compact", () => {
+  it.effect("manually compacts the active session context", () =>
+    Effect.gen(function* () {
+      requests = []
+      const session = yield* SessionV2.Service
+      const events = yield* EventV2.Service
+      const created = yield* session.create({ location })
+
+      yield* events.publish(SessionEvent.Prompted, {
+        sessionID: created.id,
+        messageID: SessionMessage.ID.create(),
+        timestamp: DateTime.makeUnsafe(0),
+        prompt: Prompt.make({ text: "Please compact this session history." }),
+        delivery: "steer",
+      })
+
+      yield* session.compact({ sessionID: created.id })
+
+      expect(requests).toHaveLength(1)
+      expect(JSON.stringify(requests[0]?.messages)).toContain("Please compact this session history.")
+      expect(yield* session.context(created.id)).toMatchObject([
+        { type: "compaction", reason: "manual", summary: "manual session summary", recent: "" },
+      ])
+    }),
+  )
+})

packages/core/test/session-compaction.test.ts

@@ -1,5 +1,55 @@
 import { expect, test } from "bun:test"
+import { LLMClient, LLMEvent, Model, type LLMRequest } from "@opencode-ai/llm"
+import { OpenAIChat } from "@opencode-ai/llm/protocols"
+import { Config } from "@opencode-ai/core/config"
+import { Database } from "@opencode-ai/core/database/database"
+import { EventV2 } from "@opencode-ai/core/event"
+import { EventTable } from "@opencode-ai/core/event/sql"
 import { SessionCompaction } from "@opencode-ai/core/session/compaction"
+import { SessionEvent } from "@opencode-ai/core/session/event"
+import { SessionMessage } from "@opencode-ai/core/session/message"
+import { SessionProjector } from "@opencode-ai/core/session/projector"
+import { SessionRunnerModel } from "@opencode-ai/core/session/runner/model"
+import { SessionTable } from "@opencode-ai/core/session/sql"
+import { SessionStore } from "@opencode-ai/core/session/store"
+import { SessionV2 } from "@opencode-ai/core/session"
+import { Project } from "@opencode-ai/core/project"
+import { ProjectTable } from "@opencode-ai/core/project/sql"
+import { AbsolutePath } from "@opencode-ai/core/schema"
+import { DateTime, Effect, Layer, Stream } from "effect"
+import { asc, eq } from "drizzle-orm"
+import { testEffect } from "./lib/effect"
+
+let requests: LLMRequest[] = []
+const model = Model.make({
+  id: "summary-model",
+  provider: "test",
+  route: OpenAIChat.route.with({ limits: { context: 10_000, output: 1_000 } }),
+})
+const client = Layer.mock(LLMClient.Service)({
+  prepare: () => Effect.die("unused"),
+  stream: (request: LLMRequest) => {
+    requests.push(request)
+    return Stream.make(LLMEvent.textDelta({ id: "summary", text: "manual summary" }))
+  },
+  generate: () => Effect.die("unused"),
+})
+const config = Layer.mock(Config.Service)({ entries: () => Effect.succeed([]) })
+const models = Layer.mock(SessionRunnerModel.Service)({ resolve: () => Effect.succeed(model) })
+const it = testEffect(
+  Layer.mergeAll(
+    Database.defaultLayer,
+    EventV2.defaultLayer,
+    SessionProjector.defaultLayer,
+    SessionStore.defaultLayer,
+    SessionCompaction.layer.pipe(
+      Layer.provide(client),
+      Layer.provide(config),
+      Layer.provide(models),
+      Layer.provide(EventV2.defaultLayer),
+    ),
+  ),
+)
 
 test("compaction describes tool media without embedding base64", () => {
   const base64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAAB"
@@ -16,3 +66,65 @@ test("compaction describes tool media without embedding base64", () => {
   expect(serialized).toBe("Image read successfully\n[Attached image/png: pixel.png]")
   expect(serialized).not.toContain(base64)
 })
+
+it.effect("manual compaction summarizes short context instead of no-op", () =>
+  Effect.gen(function* () {
+    requests = []
+    const db = (yield* Database.Service).db
+    const compaction = yield* SessionCompaction.Service
+    const store = yield* SessionStore.Service
+    const sessionID = SessionV2.ID.make("ses_manual_compaction")
+    const userMessage = {
+      id: SessionMessage.ID.create(),
+      type: "user" as const,
+      text: "Manual compaction should include this short conversation.",
+      time: { created: DateTime.makeUnsafe(0) },
+    }
+    yield* db
+      .insert(ProjectTable)
+      .values({ id: Project.ID.global, worktree: AbsolutePath.make("/project"), sandboxes: [] })
+      .onConflictDoNothing()
+      .run()
+      .pipe(Effect.orDie)
+    yield* db
+      .insert(SessionTable)
+      .values({
+        id: sessionID,
+        project_id: Project.ID.global,
+        slug: "manual-compaction",
+        directory: "/project",
+        title: "Manual compaction",
+        version: "test",
+      })
+      .run()
+      .pipe(Effect.orDie)
+
+    const session = yield* store
+      .get(sessionID)
+      .pipe(
+        Effect.flatMap((session) =>
+          session ? Effect.succeed(session) : Effect.die("manual compaction test session missing"),
+        ),
+      )
+
+    expect(yield* compaction.compactManual({ session, messages: [userMessage] })).toBe(true)
+
+    expect(requests).toHaveLength(1)
+    expect(JSON.stringify(requests[0]?.messages)).toContain("Manual compaction should include this short conversation.")
+    expect(yield* store.context(sessionID)).toMatchObject([
+      { type: "compaction", reason: "manual", summary: "manual summary", recent: "" },
+    ])
+    expect(
+      yield* db
+        .select({ type: EventTable.type })
+        .from(EventTable)
+        .where(eq(EventTable.aggregate_id, sessionID))
+        .orderBy(asc(EventTable.seq))
+        .all()
+        .pipe(Effect.orDie),
+    ).toEqual([
+      { type: EventV2.versionedType(SessionEvent.Compaction.Started.type, 1) },
+      { type: EventV2.versionedType(SessionEvent.Compaction.Ended.type, 1) },
+    ])
+  }),
+)

packages/core/test/session-prompt.test.ts

@@ -1,10 +1,12 @@
 import { describe, expect } from "bun:test"
-import { DateTime, Effect, Fiber, Layer, Stream } from "effect"
+import { DateTime, Effect, Fiber, Layer, Schema, Stream } from "effect"
 import { eq } from "drizzle-orm"
 import { Database } from "@opencode-ai/core/database/database"
 import { EventV2 } from "@opencode-ai/core/event"
 import { EventTable } from "@opencode-ai/core/event/sql"
 import { SessionEvent } from "@opencode-ai/core/session/event"
+import { ModelV2 } from "@opencode-ai/core/model"
+import { ProviderV2 } from "@opencode-ai/core/provider"
 import { Project } from "@opencode-ai/core/project"
 import { ProjectTable } from "@opencode-ai/core/project/sql"
 import { AbsolutePath } from "@opencode-ai/core/schema"
@@ -110,6 +112,21 @@ const eventCount = (type: string) =>
       ),
   )
 
+const encodeMessage = Schema.encodeSync(SessionMessage.Message)
+const assistantRow = (id: SessionMessage.ID, seq: number) => {
+  const { id: _, type, ...data } = encodeMessage(
+    SessionMessage.Assistant.make({
+      id,
+      type: "assistant",
+      agent: "build",
+      model: { id: ModelV2.ID.make("model"), providerID: ProviderV2.ID.make("provider") },
+      content: [],
+      time: { created: DateTime.makeUnsafe(0) },
+    }),
+  )
+  return { id, session_id: sessionID, type, seq, time_created: 0, data }
+}
+
 describe("SessionV2.prompt", () => {
   it.effect("exposes the execution registry", () =>
     Effect.gen(function* () {
@@ -174,30 +191,36 @@ describe("SessionV2.prompt", () => {
     }),
   )
 
-  it.effect("preserves an optional per-request system string through admission and projection", () =>
+  it.effect("commits a staged revert before admitting a new prompt", () =>
     Effect.gen(function* () {
       yield* setup
-      const { db } = yield* Database.Service
       const session = yield* SessionV2.Service
       const events = yield* EventV2.Service
+      const { db } = yield* Database.Service
 
-      const message = yield* session.prompt({
+      const boundary = yield* session.prompt({
         sessionID,
-        prompt: Prompt.make({ text: "Fix the failing tests", system: "Per-request override" }),
+        prompt: Prompt.make({ text: "boundary" }),
         resume: false,
       })
-
-      expect(message.prompt.system).toBe("Per-request override")
-      expect(yield* admitted(message.id)).toMatchObject({
-        id: message.id,
-        prompt: { text: "Fix the failing tests", system: "Per-request override" },
-      })
-
       yield* SessionInput.promoteSteers(db, events, sessionID, Number.MAX_SAFE_INTEGER)
+      const stale = SessionMessage.ID.make("msg_stale_assistant")
+      yield* db.insert(SessionMessageTable).values(assistantRow(stale, 100)).run().pipe(Effect.orDie)
+      yield* events.publish(SessionEvent.RevertEvent.Staged, {
+        sessionID,
+        timestamp: yield* DateTime.now,
+        revert: { messageID: boundary.id, files: [] },
+      })
+      expect((yield* session.get(sessionID)).revert?.messageID).toBe(boundary.id)
 
-      expect(yield* session.messages({ sessionID })).toMatchObject([
-        { id: message.id, type: "user", text: "Fix the failing tests", system: "Per-request override" },
-      ])
+      yield* session.prompt({ sessionID, prompt: Prompt.make({ text: "after revert" }), resume: false })
+
+      expect((yield* session.get(sessionID)).revert).toBeUndefined()
+      expect(
+        (yield* db.select({ id: SessionMessageTable.id }).from(SessionMessageTable).all().pipe(Effect.orDie)).map(
+          (row) => row.id,
+        ),
+      ).not.toContain(stale)
     }),
   )
 

packages/core/test/session-runner-recorded.test.ts

@@ -14,6 +14,7 @@ import { AbsolutePath } from "@opencode-ai/core/schema"
 import { SessionV2 } from "@opencode-ai/core/session"
 import { locationServiceMapLayer } from "@opencode-ai/core/location-services"
 import { Snapshot } from "@opencode-ai/core/snapshot"
+import { SessionCompaction } from "@opencode-ai/core/session/compaction"
 import { Prompt } from "@opencode-ai/core/session/prompt"
 import { SessionProjector } from "@opencode-ai/core/session/projector"
 import { SessionExecution } from "@opencode-ai/core/session/execution"
@@ -73,6 +74,7 @@ const skillGuidance = Layer.mock(SkillGuidance.Service, { load: () => Effect.suc
 const referenceGuidance = Layer.mock(ReferenceGuidance.Service, { load: () => Effect.succeed(SystemContext.empty) })
 const config = Layer.succeed(Config.Service, Config.Service.of({ entries: () => Effect.succeed([]) }))
 const runner = SessionRunnerLLM.defaultLayer.pipe(
+  Layer.provide(SessionCompaction.layer),
   Layer.provide(Snapshot.noopLayer),
   Layer.provide(Database.defaultLayer),
   Layer.provide(SessionStore.defaultLayer),

packages/core/test/session-runner-system-prompt.test.ts

@@ -0,0 +1,21 @@
+import { describe, expect, test } from "bun:test"
+import { Model } from "@opencode-ai/llm"
+import * as OpenAIChat from "@opencode-ai/llm/protocols/openai-chat"
+import { SessionRunnerSystemPrompt } from "@opencode-ai/core/session/runner/system-prompt"
+
+const prompt = (id: string) =>
+  SessionRunnerSystemPrompt.provider(Model.make({ id, provider: "test", route: OpenAIChat.route }))
+
+describe("SessionRunnerSystemPrompt", () => {
+  test("selects the legacy provider-family prompts from the model id", () => {
+    expect(prompt("gpt-5")).toContain("You are OpenCode, You and the user share the same workspace")
+    expect(prompt("gpt-4.1")).toContain("THE PROBLEM CAN NOT BE SOLVED WITHOUT EXTENSIVE INTERNET RESEARCH")
+    expect(prompt("o3")).toContain("THE PROBLEM CAN NOT BE SOLVED WITHOUT EXTENSIVE INTERNET RESEARCH")
+    expect(prompt("gpt-5-codex")).toContain("## Editing constraints")
+    expect(prompt("gemini-2.5-pro")).toContain("# Core Mandates")
+    expect(prompt("claude-sonnet-4")).toContain("# Professional objectivity")
+    expect(prompt("kimi-k2")).toContain("# Prompt and Tool Use")
+    expect(prompt("trinity")).toContain("what command should I run to list files")
+    expect(prompt("llama-3.3")).toContain("You are opencode, an interactive CLI tool")
+  })
+})

packages/core/test/session-runner.test.ts

@@ -23,6 +23,7 @@ import { locationServiceMapLayer } from "@opencode-ai/core/location-services"
 import { Snapshot } from "@opencode-ai/core/snapshot"
 import { ContextSnapshotDecodeError } from "@opencode-ai/core/session/error"
 import { SessionEvent } from "@opencode-ai/core/session/event"
+import { SessionCompaction } from "@opencode-ai/core/session/compaction"
 import { SessionInput } from "@opencode-ai/core/session/input"
 import { SessionMessage } from "@opencode-ai/core/session/message"
 import { Prompt } from "@opencode-ai/core/session/prompt"
@@ -32,6 +33,7 @@ import { SessionRunCoordinator } from "@opencode-ai/core/session/run-coordinator
 import { SessionRunner } from "@opencode-ai/core/session/runner"
 import * as SessionRunnerLLM from "@opencode-ai/core/session/runner/llm"
 import { SessionRunnerModel } from "@opencode-ai/core/session/runner/model"
+import { SessionRunnerSystemPrompt } from "@opencode-ai/core/session/runner/system-prompt"
 import { ToolRegistry } from "@opencode-ai/core/tool/registry"
 import { ToolOutputStore } from "@opencode-ai/core/tool-output-store"
 import { ApplicationTools } from "@opencode-ai/core/tool/application-tools"
@@ -96,6 +98,7 @@ const client = Layer.succeed(
   }),
 )
 const model = Model.make({ id: "fake-model", provider: "fake", route: OpenAIChat.route })
+const defaultSystem = SessionRunnerSystemPrompt.provider(model)
 const replacementModel = Model.make({ id: "replacement", provider: "fake", route: OpenAIChat.route })
 const compactModel = Model.make({
   id: "compact",
@@ -232,6 +235,7 @@ const config = Layer.succeed(
   }),
 )
 const runner = SessionRunnerLLM.layer.pipe(
+  Layer.provide(SessionCompaction.layer),
   Layer.provide(Snapshot.noopLayer),
   Layer.provide(Database.defaultLayer),
   Layer.provide(SessionStore.defaultLayer),
@@ -779,8 +783,8 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context"],
-        ["Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
       ])
       expect(requests[1]?.messages.map((message) => message.role)).toEqual(["user", "user", "system"])
       expect(requests[1]?.messages.at(-1)?.content).toEqual([{ type: "text", text: "Changed context" }])
@@ -799,6 +803,49 @@ describe("SessionRunnerLLM", () => {
     }),
   )
 
+  it.effect("uses the selected model family prompt when the agent does not override it", () =>
+    Effect.gen(function* () {
+      yield* setup
+      currentModel = Model.make({ id: "gpt-5", provider: "openai", route: OpenAIChat.route })
+      const session = yield* SessionV2.Service
+      yield* session.prompt({ sessionID, prompt: Prompt.make({ text: "First" }), resume: false })
+
+      requests.length = 0
+      response = fragmentFixture("text", "text-provider-prompt", ["Done"]).completeEvents
+      yield* session.resume(sessionID)
+
+      expect(requests.at(-1)?.system.map((part) => part.text)).toEqual([
+        expect.stringContaining("You are OpenCode, You and the user share the same workspace"),
+        "Initial context",
+      ])
+    }),
+  )
+
+  it.effect("uses the selected model family prompt when the agent system override is empty", () =>
+    Effect.gen(function* () {
+      yield* setup
+      currentModel = Model.make({ id: "gpt-5", provider: "openai", route: OpenAIChat.route })
+      const agent = yield* AgentV2.Service
+      yield* agent.transform((editor) =>
+        editor.update(AgentV2.ID.make("build"), (agent) => {
+          agent.system = ""
+          agent.mode = "primary"
+        }),
+      )
+      const session = yield* SessionV2.Service
+      yield* session.prompt({ sessionID, prompt: Prompt.make({ text: "First" }), resume: false })
+
+      requests.length = 0
+      response = fragmentFixture("text", "text-empty-agent-system", ["Done"]).completeEvents
+      yield* session.resume(sessionID)
+
+      expect(requests.at(-1)?.system.map((part) => part.text)).toEqual([
+        expect.stringContaining("You are OpenCode, You and the user share the same workspace"),
+        "Initial context",
+      ])
+    }),
+  )
+
   it.effect("includes the effective default agent system before durable context", () =>
     Effect.gen(function* () {
       yield* setup
@@ -847,36 +894,7 @@ describe("SessionRunnerLLM", () => {
     }),
   )
 
-  it.effect("appends the per-request prompt system after the agent prompt and durable baseline", () =>
-    Effect.gen(function* () {
-      yield* setup
-      const agent = yield* AgentV2.Service
-      yield* agent.transform((editor) =>
-        editor.update(AgentV2.ID.make("build"), (agent) => {
-          agent.system = "Build agent instructions"
-          agent.mode = "primary"
-        }),
-      )
-      const session = yield* SessionV2.Service
-      yield* session.prompt({
-        sessionID,
-        prompt: Prompt.make({ text: "First", system: "Per-request override" }),
-        resume: false,
-      })
-
-      requests.length = 0
-      response = fragmentFixture("text", "text-system", ["Done"]).completeEvents
-      yield* session.resume(sessionID)
-
-      expect(requests.at(-1)?.system.map((part) => part.text)).toEqual([
-        "Build agent instructions",
-        "Initial context",
-        "Per-request override",
-      ])
-    }),
-  )
-
-  it.effect("omits the per-request system part when the prompt has no system string", () =>
+  it.effect("uses only the agent prompt and durable baseline as system parts", () =>
     Effect.gen(function* () {
       yield* setup
       const agent = yield* AgentV2.Service
@@ -951,8 +969,8 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context\n\nBuild skills"],
-        ["Initial context\n\nBuild skills"],
+        [defaultSystem, "Initial context\n\nBuild skills"],
+        [defaultSystem, "Initial context\n\nBuild skills"],
       ])
       expect(systemTexts(requests[1]!)).toContainEqual(expect.stringContaining("Reviewer skills"))
     }),
@@ -985,7 +1003,7 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context\n\nBuild skills"],
+        [defaultSystem, "Initial context\n\nBuild skills"],
       ])
     }),
   )
@@ -1014,7 +1032,9 @@ describe("SessionRunnerLLM", () => {
       response = []
       yield* session.resume(sessionID)
       expect(requests.map((request) => request.model)).toEqual([model])
-      expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([["Initial context"]])
+      expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
+        [defaultSystem, "Initial context"],
+      ])
     }),
   )
 
@@ -1063,9 +1083,9 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context"],
-        ["Initial context"],
-        ["Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
       ])
       expect(requests[1]?.messages.map((message) => message.role)).toEqual(["user", "user", "system"])
       expect(requests[2]?.messages.filter((message) => message.role === "system")).toHaveLength(2)
@@ -1109,9 +1129,9 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context"],
-        ["Initial context"],
-        ["Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
       ])
     }),
   )
@@ -1146,8 +1166,8 @@ describe("SessionRunnerLLM", () => {
       yield* session.resume(sessionID)
 
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context"],
-        ["Replacement context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Replacement context"],
       ])
       yield* replaySessionProjection(sessionID)
       yield* session.prompt({ sessionID, prompt: Prompt.make({ text: "Third" }), resume: false })
@@ -1375,7 +1395,7 @@ describe("SessionRunnerLLM", () => {
       yield* session.prompt({ sessionID, prompt: Prompt.make({ text: "Third" }), resume: false })
       yield* session.resume(sessionID)
 
-      expect(requests.at(-1)?.system.map((part) => part.text)).toEqual(["Initial context"])
+      expect(requests.at(-1)?.system.map((part) => part.text)).toEqual([defaultSystem, "Initial context"])
       expect(systemTexts(requests.at(-1)!)).toContain("Changed context")
     }),
   )
@@ -1574,8 +1594,8 @@ describe("SessionRunnerLLM", () => {
 
       expect(requests.map((request) => request.model)).toEqual([model, replacementModel])
       expect(requests.map((request) => request.system.map((part) => part.text))).toEqual([
-        ["Initial context"],
-        ["Initial context"],
+        [defaultSystem, "Initial context"],
+        [defaultSystem, "Initial context"],
       ])
       expect(systemTexts(requests[1]!)).toContain("Replacement context")
     }),

packages/core/test/shell.test.ts

@@ -1,6 +1,6 @@
 import { describe, expect, test } from "bun:test"
 import path from "path"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { FSUtil } from "@opencode-ai/core/fs-util"
 import { which } from "@opencode-ai/core/util/which"
 
@@ -8,56 +8,56 @@ const withShell = async (shell: string | undefined, fn: () => void | Promise<voi
   const prev = process.env.SHELL
   if (shell === undefined) delete process.env.SHELL
   else process.env.SHELL = shell
-  Shell.acceptable.reset()
-  Shell.preferred.reset()
+  ShellSelect.acceptable.reset()
+  ShellSelect.preferred.reset()
   try {
     await fn()
   } finally {
     if (prev === undefined) delete process.env.SHELL
     else process.env.SHELL = prev
-    Shell.acceptable.reset()
-    Shell.preferred.reset()
+    ShellSelect.acceptable.reset()
+    ShellSelect.preferred.reset()
   }
 }
 
 describe("shell", () => {
   test("normalizes shell names", () => {
-    expect(Shell.name("/bin/bash")).toBe("bash")
+    expect(ShellSelect.name("/bin/bash")).toBe("bash")
     if (process.platform === "win32") {
-      expect(Shell.name("C:/tools/NU.EXE")).toBe("nu")
-      expect(Shell.name("C:/tools/PWSH.EXE")).toBe("pwsh")
+      expect(ShellSelect.name("C:/tools/NU.EXE")).toBe("nu")
+      expect(ShellSelect.name("C:/tools/PWSH.EXE")).toBe("pwsh")
     }
   })
 
   test("detects login shells", () => {
-    expect(Shell.login("/bin/bash")).toBe(true)
-    expect(Shell.login("C:/tools/pwsh.exe")).toBe(false)
+    expect(ShellSelect.login("/bin/bash")).toBe(true)
+    expect(ShellSelect.login("C:/tools/pwsh.exe")).toBe(false)
   })
 
   test("detects posix shells", () => {
-    expect(Shell.posix("/bin/bash")).toBe(true)
-    expect(Shell.posix("/bin/fish")).toBe(false)
-    expect(Shell.posix("C:/tools/pwsh.exe")).toBe(false)
+    expect(ShellSelect.posix("/bin/bash")).toBe(true)
+    expect(ShellSelect.posix("/bin/fish")).toBe(false)
+    expect(ShellSelect.posix("C:/tools/pwsh.exe")).toBe(false)
   })
 
   test("falls back when configured shell cannot be resolved", async () => {
     await withShell(undefined, async () => {
-      const preferred = Shell.preferred()
-      const acceptable = Shell.acceptable()
-      expect(Shell.preferred("opencode-missing-shell")).toBe(preferred)
-      expect(Shell.acceptable("opencode-missing-shell")).toBe(acceptable)
+      const preferred = ShellSelect.preferred()
+      const acceptable = ShellSelect.acceptable()
+      expect(ShellSelect.preferred("opencode-missing-shell")).toBe(preferred)
+      expect(ShellSelect.acceptable("opencode-missing-shell")).toBe(acceptable)
     })
   })
 
   test("falls back for terminal-only acceptable shells", () => {
-    expect(Shell.name(Shell.acceptable("fish"))).not.toBe("fish")
-    expect(Shell.name(Shell.acceptable("nu"))).not.toBe("nu")
+    expect(ShellSelect.name(ShellSelect.acceptable("fish"))).not.toBe("fish")
+    expect(ShellSelect.name(ShellSelect.acceptable("nu"))).not.toBe("nu")
   })
 
   test("builds command args per shell family", () => {
-    expect(Shell.args("/bin/sh", "echo hi", "/tmp")).toEqual(["-c", "echo hi"])
-    expect(Shell.args("/usr/bin/fish", "echo hi", "/tmp")).toEqual(["-c", "echo hi"])
-    const zsh = Shell.args("/bin/zsh", "echo hi", "/tmp")
+    expect(ShellSelect.args("/bin/sh", "echo hi", "/tmp")).toEqual(["-c", "echo hi"])
+    expect(ShellSelect.args("/usr/bin/fish", "echo hi", "/tmp")).toEqual(["-c", "echo hi"])
+    const zsh = ShellSelect.args("/bin/zsh", "echo hi", "/tmp")
     expect(zsh[0]).toBe("-l")
     expect(zsh[1]).toBe("-c")
     expect(zsh.at(-1)).toBe("/tmp")
@@ -66,34 +66,34 @@ describe("shell", () => {
   if (process.platform === "win32") {
     test("rejects blacklisted shells case-insensitively", async () => {
       await withShell("NU.EXE", async () => {
-        expect(Shell.name(Shell.acceptable())).not.toBe("nu")
+        expect(ShellSelect.name(ShellSelect.acceptable())).not.toBe("nu")
       })
     })
 
     test("normalizes Git Bash shell paths from env", async () => {
       const shell = "/cygdrive/c/Program Files/Git/bin/bash.exe"
       await withShell(shell, async () => {
-        expect(Shell.preferred()).toBe(FSUtil.windowsPath(shell))
+        expect(ShellSelect.preferred()).toBe(FSUtil.windowsPath(shell))
       })
     })
 
     test("resolves /usr/bin/bash from env to Git Bash", async () => {
-      const bash = Shell.gitbash()
+      const bash = ShellSelect.gitbash()
       if (!bash) return
       await withShell("/usr/bin/bash", async () => {
-        expect(Shell.acceptable()).toBe(bash)
-        expect(Shell.preferred()).toBe(bash)
+        expect(ShellSelect.acceptable()).toBe(bash)
+        expect(ShellSelect.preferred()).toBe(bash)
       })
     })
 
     test("resolves bare bash to Git Bash before PATH", async () => {
-      const bash = Shell.gitbash()
+      const bash = ShellSelect.gitbash()
       if (!bash) return
-      expect(Shell.acceptable("bash")).toBe(bash)
-      expect(Shell.preferred("bash")).toBe(bash)
+      expect(ShellSelect.acceptable("bash")).toBe(bash)
+      expect(ShellSelect.preferred("bash")).toBe(bash)
       await withShell("bash", async () => {
-        expect(Shell.acceptable()).toBe(bash)
-        expect(Shell.preferred()).toBe(bash)
+        expect(ShellSelect.acceptable()).toBe(bash)
+        expect(ShellSelect.preferred()).toBe(bash)
       })
     })
 
@@ -101,7 +101,7 @@ describe("shell", () => {
       const shell = which("pwsh") || which("powershell")
       if (!shell) return
       await withShell(path.win32.basename(shell), async () => {
-        expect(Shell.preferred()).toBe(shell)
+        expect(ShellSelect.preferred()).toBe(shell)
       })
     })
   }

packages/core/test/tool-bash.test.ts

@@ -1,432 +0,0 @@
-import fs from "fs/promises"
-import { realpathSync } from "node:fs"
-import path from "path"
-import { describe, expect, test } from "bun:test"
-import { Effect, Layer } from "effect"
-import { ChildProcess } from "effect/unstable/process"
-import { FSUtil } from "@opencode-ai/core/fs-util"
-import { Config } from "@opencode-ai/core/config"
-import { Location } from "@opencode-ai/core/location"
-import { LocationMutation } from "@opencode-ai/core/location-mutation"
-import { PermissionV2 } from "@opencode-ai/core/permission"
-import { AppProcess } from "@opencode-ai/core/process"
-import { AbsolutePath } from "@opencode-ai/core/schema"
-import { SessionV2 } from "@opencode-ai/core/session"
-import { BashTool } from "@opencode-ai/core/tool/bash"
-import { ToolRegistry } from "@opencode-ai/core/tool/registry"
-import { location } from "./fixture/location"
-import { tmpdir } from "./fixture/tmpdir"
-import { testEffect } from "./lib/effect"
-import { toolIdentity, executeTool, settleTool, toolDefinitions } from "./lib/tool"
-
-const sessionID = SessionV2.ID.make("ses_bash_tool_test")
-const assertions: PermissionV2.AssertInput[] = []
-const runs: Array<{
-  readonly command: string
-  readonly cwd?: string
-  readonly shell?: string | boolean
-  readonly options?: AppProcess.RunOptions
-}> = []
-let denyAction: string | undefined
-let result: AppProcess.RunResult = {
-  command: "mock",
-  exitCode: 0,
-  output: Buffer.from("hello\n"),
-  stdout: Buffer.from("hello\n"),
-  stderr: Buffer.alloc(0),
-  outputTruncated: false,
-  stdoutTruncated: false,
-  stderrTruncated: false,
-}
-let runFailure: AppProcess.AppProcessError | undefined
-let afterPermission = (_input: PermissionV2.AssertInput): Effect.Effect<void> => Effect.void
-
-const permission = Layer.succeed(
-  PermissionV2.Service,
-  PermissionV2.Service.of({
-    assert: (input) =>
-      Effect.sync(() => assertions.push(input)).pipe(
-        Effect.andThen(Effect.suspend(() => afterPermission(input))),
-        Effect.andThen(
-          input.action === denyAction ? Effect.fail(new PermissionV2.DeniedError({ rules: [] })) : Effect.void,
-        ),
-      ),
-    ask: () => Effect.die("unused"),
-    reply: () => Effect.die("unused"),
-    get: () => Effect.die("unused"),
-    forSession: () => Effect.die("unused"),
-    list: () => Effect.die("unused"),
-  }),
-)
-const appProcess = Layer.succeed(
-  AppProcess.Service,
-  AppProcess.Service.of({
-    run: (command: ChildProcess.Command, options?: AppProcess.RunOptions) =>
-      Effect.suspend(() => {
-        if (command._tag !== "StandardCommand") throw new Error("expected standard command")
-        runs.push({ command: command.command, cwd: command.options.cwd, shell: command.options.shell, options })
-        return runFailure ? Effect.fail(runFailure) : Effect.succeed(result)
-      }),
-  } as unknown as AppProcess.Interface),
-)
-const config = Layer.succeed(
-  Config.Service,
-  Config.Service.of({
-    entries: () => Effect.succeed([]),
-  }),
-)
-
-const reset = () => {
-  assertions.length = 0
-  runs.length = 0
-  denyAction = undefined
-  runFailure = undefined
-  afterPermission = () => Effect.void
-  result = {
-    command: "mock",
-    exitCode: 0,
-    output: Buffer.from("hello\n"),
-    stdout: Buffer.from("hello\n"),
-    stderr: Buffer.alloc(0),
-    outputTruncated: false,
-    stdoutTruncated: false,
-    stderrTruncated: false,
-  }
-}
-
-const withTool = <A, E, R>(
-  directory: string,
-  body: (registry: ToolRegistry.Interface) => Effect.Effect<A, E, R>,
-  processLayer: Layer.Layer<AppProcess.Service> = appProcess,
-) => {
-  const filesystem = FSUtil.defaultLayer
-  const activeLocation = Layer.succeed(
-    Location.Service,
-    Location.Service.of(location({ directory: AbsolutePath.make(directory) })),
-  )
-  const mutation = LocationMutation.layer.pipe(Layer.provide(filesystem), Layer.provide(activeLocation))
-  const registry = ToolRegistry.defaultLayer.pipe(Layer.provide(permission))
-  const bash = BashTool.layer.pipe(
-    Layer.provide(registry),
-    Layer.provide(permission),
-    Layer.provide(mutation),
-    Layer.provide(filesystem),
-    Layer.provide(processLayer),
-    Layer.provide(config),
-  )
-  return Effect.gen(function* () {
-    return yield* body(yield* ToolRegistry.Service)
-  }).pipe(Effect.provide(Layer.mergeAll(registry, bash)))
-}
-
-const call = (input: typeof BashTool.Input.Type, id = "call-bash") => ({
-  sessionID,
-  ...toolIdentity,
-  call: { type: "tool-call" as const, id, name: "bash", input },
-})
-
-const it = testEffect(Layer.empty)
-
-describe("BashTool", () => {
-  it.live("registers and returns structured successful output from the active Location", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        return withTool(tmp.path, (registry) =>
-          Effect.gen(function* () {
-            const definitions = yield* toolDefinitions(registry)
-            expect(definitions.map((tool) => tool.name)).toEqual(["bash"])
-            expect(definitions[0]?.inputSchema).not.toHaveProperty("properties.background")
-            expect(definitions[0]?.inputSchema).not.toHaveProperty("properties.description")
-            expect(definitions[0]?.outputSchema).not.toHaveProperty("properties.output")
-            expect(definitions[0]?.outputSchema).not.toHaveProperty("properties.command")
-            expect(definitions[0]?.outputSchema).not.toHaveProperty("properties.cwd")
-            expect(yield* toolDefinitions(registry, [{ action: "bash", resource: "*", effect: "deny" }])).toEqual([])
-            expect(yield* settleTool(registry, call({ command: "pwd" }))).toEqual({
-              result: {
-                type: "content",
-                value: [
-                  { type: "text", text: "hello\n" },
-                  { type: "text", text: "Command exited with code 0." },
-                ],
-              },
-              output: {
-                structured: {
-                  exit: 0,
-                  truncated: false,
-                },
-                content: [
-                  { type: "text", text: "hello\n" },
-                  { type: "text", text: "Command exited with code 0." },
-                ],
-              },
-            })
-            expect(runs).toMatchObject([{ command: "pwd", cwd: realpathSync(tmp.path) }])
-            expect(runs[0]?.options).toMatchObject({
-              combineOutput: true,
-              maxOutputBytes: BashTool.MAX_CAPTURE_BYTES,
-            })
-            expect(assertions).toMatchObject([{ sessionID, action: "bash", resources: ["pwd"], save: ["pwd"] }])
-          }),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-
-  it.live("resolves a relative workdir from the active Location", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        return Effect.promise(() => fs.mkdir(path.join(tmp.path, "src"))).pipe(
-          Effect.andThen(
-            withTool(tmp.path, (registry) => executeTool(registry, call({ command: "pwd", workdir: "src" }))),
-          ),
-          Effect.andThen(
-            Effect.sync(() => expect(runs).toMatchObject([{ cwd: realpathSync(path.join(tmp.path, "src")) }])),
-          ),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-
-  it.live("rejects a workdir that stops being a directory during approval", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        const workdir = path.join(tmp.path, "src")
-        afterPermission = (input) =>
-          input.action === "bash"
-            ? Effect.promise(async () => {
-                await fs.rm(workdir, { recursive: true })
-                await fs.writeFile(workdir, "not a directory")
-              }).pipe(Effect.orDie)
-            : Effect.void
-        return Effect.promise(() => fs.mkdir(workdir)).pipe(
-          Effect.andThen(
-            withTool(tmp.path, (registry) => executeTool(registry, call({ command: "pwd", workdir: "src" }))),
-          ),
-          Effect.andThen(
-            Effect.sync(() => {
-              expect(runs).toEqual([])
-              expect(assertions.map((input) => input.action)).toEqual(["bash"])
-            }),
-          ),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-
-  if (process.platform !== "win32") {
-    it.live("executes a real shell command through AppProcess", () =>
-      Effect.acquireUseRelease(
-        Effect.promise(() => tmpdir()),
-        (tmp) => {
-          reset()
-          return withTool(
-            tmp.path,
-            (registry) => settleTool(registry, call({ command: "printf core-bash" })),
-            AppProcess.defaultLayer,
-          ).pipe(
-            Effect.andThen((settled) =>
-              Effect.sync(() => {
-                expect(settled.result).toEqual({
-                  type: "content",
-                  value: [
-                    { type: "text", text: "core-bash" },
-                    { type: "text", text: "Command exited with code 0." },
-                  ],
-                })
-                expect(settled.output?.structured).toMatchObject({
-                  exit: 0,
-                })
-                expect(settled.output?.structured).not.toHaveProperty("output")
-              }),
-            ),
-          )
-        },
-        (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-      ),
-    )
-  }
-
-  it.live("approves an explicit external workdir before bash execution", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
-      ([active, outside]) => {
-        reset()
-        return withTool(active.path, (registry) =>
-          executeTool(registry, call({ command: "pwd", workdir: outside.path })),
-        ).pipe(
-          Effect.andThen(
-            Effect.sync(() => {
-              expect(assertions.map((item) => item.action)).toEqual(["external_directory", "bash"])
-              expect(assertions[0]).toMatchObject({
-                resources: [path.join(realpathSync(outside.path), "*").replaceAll("\\", "/")],
-              })
-              expect(runs).toHaveLength(1)
-            }),
-          ),
-        )
-      },
-      ([active, outside]) =>
-        Effect.promise(() =>
-          Promise.all([active[Symbol.asyncDispose](), outside[Symbol.asyncDispose]()]).then(() => undefined),
-        ),
-    ),
-  )
-
-  it.live("does not execute after external-directory or bash denial", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
-      ([active, outside]) =>
-        Effect.gen(function* () {
-          reset()
-          denyAction = "external_directory"
-          yield* withTool(active.path, (registry) =>
-            executeTool(registry, call({ command: "pwd", workdir: outside.path })),
-          )
-          expect(assertions.map((item) => item.action)).toEqual(["external_directory"])
-          expect(runs).toEqual([])
-
-          reset()
-          denyAction = "bash"
-          yield* withTool(active.path, (registry) => executeTool(registry, call({ command: "pwd" })))
-          expect(assertions.map((item) => item.action)).toEqual(["bash"])
-          expect(runs).toEqual([])
-        }),
-      ([active, outside]) =>
-        Effect.promise(() =>
-          Promise.all([active[Symbol.asyncDispose](), outside[Symbol.asyncDispose]()]).then(() => undefined),
-        ),
-    ),
-  )
-
-  it.live("reports external command arguments as advisory warnings without enforcing approval", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
-      ([active, outside]) => {
-        reset()
-        denyAction = "external_directory"
-        const target = path.join(outside.path, "secret.txt")
-        return withTool(active.path, (registry) => settleTool(registry, call({ command: `cat ${target}` }))).pipe(
-          Effect.andThen((settled) =>
-            Effect.sync(() => {
-              expect(assertions.map((item) => item.action)).toEqual(["bash"])
-              expect(runs).toHaveLength(1)
-              expect(settled.output?.structured).toMatchObject({
-                truncated: false,
-              })
-              expect(settled.output?.structured).not.toHaveProperty("warnings")
-              expect(settled.output?.content[1]).toMatchObject({
-                type: "text",
-                text: expect.stringContaining("Warnings:"),
-              })
-            }),
-          ),
-        )
-      },
-      ([active, outside]) =>
-        Effect.promise(() =>
-          Promise.all([active[Symbol.asyncDispose](), outside[Symbol.asyncDispose]()]).then(() => undefined),
-        ),
-    ),
-  )
-
-  it.live("keeps non-zero exits useful", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        result = { ...result, exitCode: 7, output: Buffer.from("HEAD full output TAIL") }
-        return withTool(tmp.path, (registry) => settleTool(registry, call({ command: "false" }, "call-overflow"))).pipe(
-          Effect.andThen((settled) =>
-            Effect.sync(() => {
-              expect(settled.output?.content[1]).toMatchObject({
-                type: "text",
-                text: expect.stringContaining("Command exited with code 7"),
-              })
-              expect(settled.output?.structured).toMatchObject({
-                exit: 7,
-                truncated: false,
-              })
-              expect(settled.output?.content[0]).toEqual({ type: "text", text: "HEAD full output TAIL" })
-            }),
-          ),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-
-  it.live("surfaces bounded process-capture truncation", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        result = { ...result, outputTruncated: true }
-        return withTool(tmp.path, (registry) => settleTool(registry, call({ command: "verbose" }))).pipe(
-          Effect.andThen((settled) =>
-            Effect.sync(() => {
-              expect(settled.output?.structured).toMatchObject({ truncated: true })
-              expect(settled.output?.content[0]).toMatchObject({
-                type: "text",
-                text: expect.stringContaining("output capture truncated"),
-              })
-              expect(settled.output?.structured).not.toHaveProperty("resource")
-            }),
-          ),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-
-  it.live("returns a useful timeout settlement", () =>
-    Effect.acquireUseRelease(
-      Effect.promise(() => tmpdir()),
-      (tmp) => {
-        reset()
-        runFailure = new AppProcess.AppProcessError({ command: "sleep", cause: new Error("Timed out") })
-        return withTool(tmp.path, (registry) => settleTool(registry, call({ command: "sleep 60", timeout: 10 }))).pipe(
-          Effect.andThen((settled) =>
-            Effect.sync(() => {
-              expect(settled.output?.content[1]).toMatchObject({
-                type: "text",
-                text: expect.stringContaining("Command timed out"),
-              })
-              expect(settled.output?.structured).toMatchObject({
-                timeout: true,
-                truncated: false,
-              })
-            }),
-          ),
-        )
-      },
-      (tmp) => Effect.promise(() => tmp[Symbol.asyncDispose]()),
-    ),
-  )
-})
-
-test("keeps locked deferred parity TODOs visible", async () => {
-  const source = await fs.readFile(new URL("../src/tool/bash.ts", import.meta.url), "utf8")
-  for (const todo of [
-    "Port tree-sitter bash / PowerShell parser-based approval reduction.",
-    "Port BashArity reusable command-prefix approvals.",
-    "Replace token-based command-argument external-directory advisories with parser-based detection.",
-    "Restore PowerShell and cmd-specific invocation/path handling on Windows.",
-    "Add plugin shell.env environment augmentation once V2 plugin hooks exist.",
-    "Add durable/live progress metadata streaming for long-running commands once V2 tool invocation progress context is wired.",
-    "Persist background job status and define restart recovery before exposing remote observation.",
-    "Revisit process-group cleanup and platform coverage with shell-specific tests if current AppProcess semantics do not fully cover it.",
-    "Revisit binary output handling if stdout/stderr decoding is text-only.",
-    "Stream full shell output into managed storage while retaining only a bounded in-memory preview.",
-  ]) {
-    expect(source).toContain(`TODO: ${todo}`)
-  }
-})

packages/core/test/tool-shell.test.ts

@@ -0,0 +1,366 @@
+import fs from "fs/promises"
+import { realpathSync } from "node:fs"
+import path from "path"
+import { describe, expect, test } from "bun:test"
+import { Effect, Layer } from "effect"
+import { Config } from "@opencode-ai/core/config"
+import { EventV2 } from "@opencode-ai/core/event"
+import { FSUtil } from "@opencode-ai/core/fs-util"
+import { Global } from "@opencode-ai/core/global"
+import { Location } from "@opencode-ai/core/location"
+import { LocationMutation } from "@opencode-ai/core/location-mutation"
+import { PermissionV2 } from "@opencode-ai/core/permission"
+import { AppProcess } from "@opencode-ai/core/process"
+import { Project } from "@opencode-ai/core/project"
+import { AbsolutePath } from "@opencode-ai/core/schema"
+import { SessionV2 } from "@opencode-ai/core/session"
+import { Shell } from "@opencode-ai/core/shell"
+import { ShellTool } from "@opencode-ai/core/tool/shell"
+import { ToolRegistry } from "@opencode-ai/core/tool/registry"
+import { tmpdir } from "./fixture/tmpdir"
+import { testEffect } from "./lib/effect"
+import { toolIdentity, executeTool, settleTool, toolDefinitions } from "./lib/tool"
+
+const sessionID = SessionV2.ID.make("ses_shell_tool_test")
+const assertions: PermissionV2.AssertInput[] = []
+let denyAction: string | undefined
+let afterPermission = (_input: PermissionV2.AssertInput): Effect.Effect<void> => Effect.void
+
+const permission = Layer.succeed(
+  PermissionV2.Service,
+  PermissionV2.Service.of({
+    assert: (input) =>
+      Effect.sync(() => assertions.push(input)).pipe(
+        Effect.andThen(Effect.suspend(() => afterPermission(input))),
+        Effect.andThen(
+          input.action === denyAction ? Effect.fail(new PermissionV2.DeniedError({ rules: [] })) : Effect.void,
+        ),
+      ),
+    ask: () => Effect.die("unused"),
+    reply: () => Effect.die("unused"),
+    get: () => Effect.die("unused"),
+    forSession: () => Effect.die("unused"),
+    list: () => Effect.die("unused"),
+  }),
+)
+
+const reset = () => {
+  assertions.length = 0
+  denyAction = undefined
+  afterPermission = () => Effect.void
+}
+
+const withTool = <A, E, R>(
+  data: string,
+  directory: string,
+  body: (registry: ToolRegistry.Interface) => Effect.Effect<A, E, R>,
+) => {
+  const filesystem = FSUtil.defaultLayer
+  const location = Location.layer(Location.Ref.make({ directory: AbsolutePath.make(directory) })).pipe(
+    Layer.provide(Project.defaultLayer),
+  )
+  const global = Global.layerWith({ data, config: path.join(data, "config") })
+  const mutation = LocationMutation.layer.pipe(Layer.provide(filesystem), Layer.provide(location))
+  const registry = ToolRegistry.defaultLayer.pipe(Layer.provide(permission))
+  const shellService = Shell.layer.pipe(
+    Layer.provide(EventV2.defaultLayer),
+    Layer.provide(location),
+    Layer.provide(Config.locationLayer.pipe(Layer.provide(location), Layer.provide(filesystem), Layer.provide(global))),
+    Layer.provide(global),
+    Layer.provide(filesystem),
+    Layer.provide(AppProcess.defaultLayer),
+  )
+  const shell = ShellTool.layer.pipe(
+    Layer.provide(registry),
+    Layer.provide(permission),
+    Layer.provide(mutation),
+    Layer.provide(filesystem),
+    Layer.provide(shellService),
+  )
+  return Effect.gen(function* () {
+    return yield* body(yield* ToolRegistry.Service)
+  }).pipe(Effect.provide(Layer.mergeAll(registry, shell, filesystem)))
+}
+
+const call = (input: typeof ShellTool.Input.Type, id = "call-shell") => ({
+  sessionID,
+  ...toolIdentity,
+  call: { type: "tool-call" as const, id, name: "shell", input },
+})
+
+const it = testEffect(Layer.empty)
+
+describe("ShellTool", () => {
+  it.live("registers and returns real successful output from the active Location", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        return withTool(data.path, tmp.path, (registry) =>
+          Effect.gen(function* () {
+            const definitions = yield* toolDefinitions(registry)
+            expect(definitions.map((tool) => tool.name)).toEqual(["shell"])
+            expect(definitions[0]?.outputSchema).not.toHaveProperty("properties.output")
+            expect(yield* toolDefinitions(registry, [{ action: "shell", resource: "*", effect: "deny" }])).toEqual([])
+
+            const settled = yield* settleTool(registry, call({ command: "printf hello" }))
+            expect(settled.output?.structured).toMatchObject({ exit: 0, truncated: false })
+            expect(settled.output?.content[0]).toEqual({ type: "text", text: "hello" })
+            expect(settled.output?.content[1]).toMatchObject({
+              type: "text",
+              text: expect.stringContaining("Command exited with code 0."),
+            })
+            expect(assertions).toMatchObject([{ sessionID, action: "shell", resources: ["printf hello"] }])
+          }),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("resolves a relative workdir from the active Location", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        return Effect.promise(() => fs.mkdir(path.join(tmp.path, "src"))).pipe(
+          Effect.andThen(
+            withTool(data.path, tmp.path, (registry) => settleTool(registry, call({ command: "pwd", workdir: "src" }))),
+          ),
+          Effect.andThen((settled) =>
+            Effect.sync(() =>
+              expect(settled.output?.content[0]).toMatchObject({
+                type: "text",
+                text: expect.stringContaining(realpathSync(path.join(tmp.path, "src"))),
+              }),
+            ),
+          ),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("rejects a workdir that stops being a directory during approval", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        const workdir = path.join(tmp.path, "src")
+        afterPermission = (input) =>
+          input.action === "shell"
+            ? Effect.promise(async () => {
+                await fs.rm(workdir, { recursive: true })
+                await fs.writeFile(workdir, "not a directory")
+              }).pipe(Effect.orDie)
+            : Effect.void
+        return Effect.promise(() => fs.mkdir(workdir)).pipe(
+          Effect.andThen(
+            withTool(data.path, tmp.path, (registry) =>
+              executeTool(registry, call({ command: "pwd", workdir: "src" })),
+            ),
+          ),
+          Effect.andThen(Effect.sync(() => expect(assertions.map((input) => input.action)).toEqual(["shell"]))),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("approves an explicit external workdir before shell execution", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir(), tmpdir()])),
+      ([data, active, outside]) => {
+        reset()
+        return withTool(data.path, active.path, (registry) =>
+          executeTool(registry, call({ command: "pwd", workdir: outside.path })),
+        ).pipe(
+          Effect.andThen(
+            Effect.sync(() => {
+              expect(assertions.map((item) => item.action)).toEqual(["external_directory", "shell"])
+              expect(assertions[0]).toMatchObject({
+                resources: [path.join(realpathSync(outside.path), "*").replaceAll("\\", "/")],
+              })
+            }),
+          ),
+        )
+      },
+      ([data, active, outside]) =>
+        Effect.promise(() =>
+          Promise.all([
+            data[Symbol.asyncDispose](),
+            active[Symbol.asyncDispose](),
+            outside[Symbol.asyncDispose](),
+          ]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("does not execute after external-directory or shell denial", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir(), tmpdir()])),
+      ([data, active, outside]) =>
+        Effect.gen(function* () {
+          reset()
+          denyAction = "external_directory"
+          yield* withTool(data.path, active.path, (registry) =>
+            executeTool(registry, call({ command: "pwd", workdir: outside.path })),
+          )
+          expect(assertions.map((item) => item.action)).toEqual(["external_directory"])
+
+          reset()
+          denyAction = "shell"
+          yield* withTool(data.path, active.path, (registry) => executeTool(registry, call({ command: "pwd" })))
+          expect(assertions.map((item) => item.action)).toEqual(["shell"])
+        }),
+      ([data, active, outside]) =>
+        Effect.promise(() =>
+          Promise.all([
+            data[Symbol.asyncDispose](),
+            active[Symbol.asyncDispose](),
+            outside[Symbol.asyncDispose](),
+          ]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("reports external command arguments as advisory warnings without enforcing approval", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir(), tmpdir()])),
+      ([data, active, outside]) => {
+        reset()
+        denyAction = "external_directory"
+        const target = path.join(outside.path, "secret.txt")
+        return withTool(data.path, active.path, (registry) =>
+          settleTool(registry, call({ command: `cat ${target}` })),
+        ).pipe(
+          Effect.andThen((settled) =>
+            Effect.sync(() => {
+              expect(assertions.map((item) => item.action)).toEqual(["shell"])
+              expect(settled.output?.structured).not.toHaveProperty("warnings")
+              expect(settled.output?.content[1]).toMatchObject({
+                type: "text",
+                text: expect.stringContaining("Warnings:"),
+              })
+            }),
+          ),
+        )
+      },
+      ([data, active, outside]) =>
+        Effect.promise(() =>
+          Promise.all([
+            data[Symbol.asyncDispose](),
+            active[Symbol.asyncDispose](),
+            outside[Symbol.asyncDispose](),
+          ]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("keeps non-zero exits useful", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        return withTool(data.path, tmp.path, (registry) =>
+          settleTool(registry, call({ command: "printf body && exit 7" }, "call-nonzero")),
+        ).pipe(
+          Effect.andThen((settled) =>
+            Effect.sync(() => {
+              expect(settled.output?.structured).toMatchObject({ exit: 7, truncated: false })
+              expect(settled.output?.content[0]).toEqual({ type: "text", text: "body" })
+              expect(settled.output?.content[1]).toMatchObject({
+                type: "text",
+                text: expect.stringContaining("Command exited with code 7"),
+              })
+            }),
+          ),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("truncates the model view and points at the saved output file when output overflows", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        const bytes = ShellTool.MAX_CAPTURE_BYTES + 1024
+        return withTool(data.path, tmp.path, (registry) =>
+          settleTool(registry, call({ command: `head -c ${bytes} /dev/zero | tr '\\0' 'x'` }, "call-overflow")),
+        ).pipe(
+          Effect.andThen((settled) =>
+            Effect.sync(() => {
+              expect(settled.output?.structured).toMatchObject({ exit: 0, truncated: true })
+              expect(settled.output?.content[0]).toMatchObject({
+                type: "text",
+                text: expect.stringContaining("output truncated; full output saved to:"),
+              })
+            }),
+          ),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+
+  it.live("returns a useful timeout settlement", () =>
+    Effect.acquireUseRelease(
+      Effect.promise(() => Promise.all([tmpdir(), tmpdir()])),
+      ([data, tmp]) => {
+        reset()
+        return withTool(data.path, tmp.path, (registry) =>
+          settleTool(registry, call({ command: "sleep 60", timeout: 50 })),
+        ).pipe(
+          Effect.andThen((settled) =>
+            Effect.sync(() => {
+              expect(settled.output?.structured).toMatchObject({ timeout: true, truncated: false })
+              expect(settled.output?.content[1]).toMatchObject({
+                type: "text",
+                text: expect.stringContaining("Command timed out"),
+              })
+            }),
+          ),
+        )
+      },
+      ([data, tmp]) =>
+        Effect.promise(() =>
+          Promise.all([data[Symbol.asyncDispose](), tmp[Symbol.asyncDispose]()]).then(() => undefined),
+        ),
+    ),
+  )
+})
+
+test("keeps locked deferred parity TODOs visible", async () => {
+  const source = await fs.readFile(new URL("../src/tool/shell.ts", import.meta.url), "utf8")
+  for (const todo of [
+    "Port tree-sitter bash / PowerShell parser-based approval reduction.",
+    "Port BashArity reusable command-prefix approvals.",
+    "Replace token-based command-argument external-directory advisories with parser-based detection.",
+    "Restore PowerShell and cmd-specific invocation/path handling on Windows.",
+    "Add plugin shell.env environment augmentation once V2 plugin hooks exist.",
+    "Add durable/live progress metadata streaming for long-running commands once V2 tool invocation progress context is wired.",
+    "Persist background job status and define restart recovery before exposing remote observation.",
+    "Revisit process-group cleanup and platform coverage with shell-specific tests if current AppProcess semantics do not fully cover it.",
+    "Revisit binary output handling if stdout/stderr decoding is text-only.",
+    "Stream full shell output into managed storage while retaining only a bounded in-memory preview.",
+  ]) {
+    expect(source).toContain(`TODO: ${todo}`)
+  }
+})

packages/opencode/src/server/routes/instance/httpapi/handlers/pty.ts

@@ -9,7 +9,7 @@ import { PtyTicket } from "@opencode-ai/core/pty/ticket"
 import { LocationServiceMap, locationServiceMapLayer } from "@opencode-ai/core/location-services"
 import { Location } from "@opencode-ai/core/location"
 import { AbsolutePath } from "@opencode-ai/core/schema"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { CorsConfig, isAllowedRequestOrigin, type CorsOptions } from "@opencode-ai/server/cors"
 import {
   PTY_CONNECT_TICKET_QUERY,
@@ -58,7 +58,7 @@ export const ptyHandlers = HttpApiBuilder.group(InstanceHttpApi, "pty", (handler
     })
 
     const shells = Effect.fn("PtyHttpApi.shells")(function* () {
-      return yield* Effect.promise(() => Shell.list())
+      return yield* Effect.promise(() => ShellSelect.list())
     })
 
     const list = Effect.fn("PtyHttpApi.list")(function* () {

packages/opencode/src/session/prompt.ts

@@ -35,7 +35,7 @@ import { Tool } from "@/tool/tool"
 import { Permission } from "@/permission"
 import { SessionStatus } from "./status"
 import { LLM } from "./llm"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { ShellID } from "@/tool/shell/id"
 import { FSUtil } from "@opencode-ai/core/fs-util"
 import { Truncate } from "@/tool/truncate"
@@ -520,8 +520,8 @@ export const layer = Layer.effect(
           }).pipe(Effect.ensuring(markReady))
 
           const cfg = yield* config.get()
-          const sh = Shell.preferred(cfg.shell)
-          const args = Shell.args(sh, input.command, cwd)
+          const sh = ShellSelect.preferred(cfg.shell)
+          const args = ShellSelect.args(sh, input.command, cwd)
           let output = ""
           let aborted = false
 
@@ -1396,7 +1396,7 @@ export const layer = Layer.effect(
       const shellMatches = ConfigMarkdown.shell(template)
       if (shellMatches.length > 0) {
         const cfg = yield* config.get()
-        const sh = Shell.preferred(cfg.shell)
+        const sh = ShellSelect.preferred(cfg.shell)
         const results = yield* Effect.promise(() =>
           Promise.all(
             shellMatches.map(async ([, cmd]) => (await Process.text([cmd], { shell: sh, nothrow: true })).text),

packages/opencode/src/tool/shell.ts

@@ -12,7 +12,7 @@ import { FSUtil } from "@opencode-ai/core/fs-util"
 import { fileURLToPath } from "url"
 import { Config } from "@/config/config"
 import { RuntimeFlags } from "@/effect/runtime-flags"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { ShellID } from "./shell/id"
 
 import * as Truncate from "./truncate"
@@ -291,7 +291,7 @@ const ask = Effect.fn("ShellTool.ask")(function* (ctx: Tool.Context, scan: Scan,
 })
 
 function cmd(shell: string, command: string, cwd: string, env: NodeJS.ProcessEnv) {
-  if (process.platform === "win32" && Shell.ps(shell)) {
+  if (process.platform === "win32" && ShellSelect.ps(shell)) {
     return ChildProcess.make(shell, ["-NoLogo", "-NoProfile", "-NonInteractive", "-Command", command], {
       cwd,
       env,
@@ -357,7 +357,7 @@ export const ShellTool = Tool.define(
 
     const resolvePath = Effect.fn("ShellTool.resolvePath")(function* (text: string, root: string, shell: string) {
       if (process.platform === "win32") {
-        if (Shell.posix(shell) && text.startsWith("/") && FSUtil.windowsPath(text) === text) {
+        if (ShellSelect.posix(shell) && text.startsWith("/") && FSUtil.windowsPath(text) === text) {
           const file = yield* cygpath(shell, text)
           if (file) return file
         }
@@ -387,7 +387,7 @@ export const ShellTool = Tool.define(
         patterns: new Set<string>(),
         always: new Set<string>(),
       }
-      const shellKind = ShellID.toKind(Shell.name(shell))
+      const shellKind = ShellID.toKind(ShellSelect.name(shell))
 
       for (const node of commands(root)) {
         const command = parts(node)
@@ -597,8 +597,8 @@ export const ShellTool = Tool.define(
     return () =>
       Effect.gen(function* () {
         const cfg = yield* config.get()
-        const shell = Shell.acceptable(cfg.shell)
-        const name = Shell.name(shell)
+        const shell = ShellSelect.acceptable(cfg.shell)
+        const name = ShellSelect.name(shell)
         const limits = yield* trunc.limits()
         const prompt = ShellPrompt.render(name, process.platform, limits, defaultTimeoutMs)
         yield* Effect.logInfo("shell tool using shell", { shell })
@@ -616,7 +616,7 @@ export const ShellTool = Tool.define(
                 throw new Error(`Invalid timeout value: ${params.timeout}. Timeout must be a positive number.`)
               }
               const timeout = params.timeout ?? defaultTimeoutMs
-              const ps = Shell.ps(shell)
+              const ps = ShellSelect.ps(shell)
               yield* Effect.scoped(
                 Effect.gen(function* () {
                   const tree = yield* Effect.acquireRelease(parse(params.command, ps), (tree) =>

packages/opencode/test/session/prompt.test.ts

@@ -43,7 +43,7 @@ import { SessionV2 } from "@opencode-ai/core/session"
 import { SessionExecution } from "@opencode-ai/core/session/execution"
 import { Skill } from "../../src/skill"
 import { SystemPrompt } from "../../src/session/system"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { Snapshot } from "../../src/snapshot"
 import { ToolRegistry } from "@/tool/registry"
 import { Truncate } from "@/tool/truncate"
@@ -77,7 +77,7 @@ function withSh<A, E, R>(fx: () => Effect.Effect<A, E, R>) {
     Effect.sync(() => {
       const prev = process.env.SHELL
       process.env.SHELL = "/bin/sh"
-      Shell.preferred.reset()
+      ShellSelect.preferred.reset()
       return prev
     }),
     () => fx(),
@@ -85,7 +85,7 @@ function withSh<A, E, R>(fx: () => Effect.Effect<A, E, R>) {
       Effect.sync(() => {
         if (prev === undefined) delete process.env.SHELL
         else process.env.SHELL = prev
-        Shell.preferred.reset()
+        ShellSelect.preferred.reset()
       }),
   )
 }

packages/opencode/test/tool/shell.test.ts

@@ -5,7 +5,7 @@ import type * as Scope from "effect/Scope"
 import os from "os"
 import path from "path"
 import { Config } from "@/config/config"
-import { Shell } from "@opencode-ai/core/shell"
+import { ShellSelect } from "@opencode-ai/core/shell/select"
 import { ShellTool } from "../../src/tool/shell"
 import { Filesystem } from "@/util/filesystem"
 import { provideInstance, testInstanceStoreLayer, tmpdirScoped } from "../fixture/fixture"
@@ -77,25 +77,25 @@ const ctx = {
   ask: () => Effect.void,
 }
 
-Shell.acceptable.reset()
+ShellSelect.acceptable.reset()
 const quote = (text: string) => `"${text}"`
 const squote = (text: string) => `'${text}'`
 const projectRoot = path.join(__dirname, "../..")
 const bin = quote(process.execPath.replaceAll("\\", "/"))
 const bash = (() => {
-  const shell = Shell.acceptable()
-  if (Shell.name(shell) === "bash") return shell
-  return Shell.gitbash()
+  const shell = ShellSelect.acceptable()
+  if (ShellSelect.name(shell) === "bash") return shell
+  return ShellSelect.gitbash()
 })()
 const shells = (() => {
   if (process.platform !== "win32") {
-    const shell = Shell.acceptable()
-    return [{ label: Shell.name(shell), shell }]
+    const shell = ShellSelect.acceptable()
+    return [{ label: ShellSelect.name(shell), shell }]
   }
 
   const list = [bash, Bun.which("pwsh"), Bun.which("powershell"), process.env.COMSPEC || Bun.which("cmd.exe")]
     .filter((shell): shell is string => Boolean(shell))
-    .map((shell) => ({ label: Shell.name(shell), shell }))
+    .map((shell) => ({ label: ShellSelect.name(shell), shell }))
 
   return list.filter(
     (item, i) => list.findIndex((other) => other.shell.toLowerCase() === item.shell.toLowerCase()) === i,
@@ -105,7 +105,7 @@ const PS = new Set(["pwsh", "powershell"])
 const ps = shells.filter((item) => PS.has(item.label))
 const cmdShell = shells.find((item) => item.label === "cmd")
 
-const sh = () => Shell.name(Shell.acceptable())
+const sh = () => ShellSelect.name(ShellSelect.acceptable())
 const evalarg = (text: string) => (sh() === "cmd" ? quote(text) : squote(text))
 
 const fill = (mode: "lines" | "bytes", n: number) => {
@@ -133,8 +133,8 @@ const withShell = <A, E, R>(item: { label: string; shell: string }, self: Effect
     Effect.sync(() => {
       const prev = process.env.SHELL
       process.env.SHELL = item.shell
-      Shell.acceptable.reset()
-      Shell.preferred.reset()
+      ShellSelect.acceptable.reset()
+      ShellSelect.preferred.reset()
       return prev
     }),
     () => self,
@@ -142,8 +142,8 @@ const withShell = <A, E, R>(item: { label: string; shell: string }, self: Effect
       Effect.sync(() => {
         if (prev === undefined) delete process.env.SHELL
         else process.env.SHELL = prev
-        Shell.acceptable.reset()
-        Shell.preferred.reset()
+        ShellSelect.acceptable.reset()
+        ShellSelect.preferred.reset()
       }),
   )
 
@@ -196,7 +196,7 @@ describe("tool.shell", () => {
         tmp,
         Effect.gen(function* () {
           const bash = yield* initBash()
-          const fallback = Shell.name(Shell.acceptable("fish"))
+          const fallback = ShellSelect.name(ShellSelect.acceptable("fish"))
           expect(fallback).not.toBe("fish")
           expect(bash.description).toContain(fallback)
 

packages/plugin/src/v2/effect/context.ts

@@ -6,7 +6,6 @@ import type { CommandHooks } from "./command.js"
 import type { IntegrationHooks } from "./integration.js"
 import type { PluginDomain } from "./plugin.js"
 import type { ReferenceHooks } from "./reference.js"
-import type { SessionDomain } from "./session.js"
 import type { SkillHooks } from "./skill.js"
 import type { Reload } from "./registration.js"
 
@@ -19,6 +18,5 @@ export interface PluginContext {
   readonly integration: IntegrationHooks & Reload
   readonly plugin: PluginDomain
   readonly reference: ReferenceHooks & Reload
-  readonly session: SessionDomain
   readonly skill: SkillHooks & Reload
 }

packages/plugin/src/v2/effect/session.ts

@@ -1,30 +0,0 @@
-import type { Effect } from "effect"
-import type { PromptInput, SessionInputAdmitted, SessionMessage, SessionV2Info } from "@opencode-ai/sdk/v2/types"
-
-export interface SessionDomain {
-  readonly create: (input: {
-    readonly id?: string
-    readonly parentID?: string
-    readonly title?: string
-    readonly agent?: string
-    readonly model?: SessionV2Info["model"]
-  }) => Effect.Effect<SessionV2Info, unknown>
-  readonly get: (sessionID: string) => Effect.Effect<SessionV2Info, unknown>
-  readonly messages: (input: {
-    readonly sessionID: string
-    readonly limit?: number
-    readonly order?: "asc" | "desc"
-    readonly cursor?: { readonly id: string; readonly direction: "previous" | "next" }
-  }) => Effect.Effect<ReadonlyArray<SessionMessage>, unknown>
-  readonly context: (sessionID: string) => Effect.Effect<ReadonlyArray<SessionMessage>, unknown>
-  readonly prompt: (input: {
-    readonly id?: string
-    readonly sessionID: string
-    readonly prompt: PromptInput
-    readonly delivery?: "steer" | "queue"
-    readonly resume?: boolean
-  }) => Effect.Effect<SessionInputAdmitted, unknown>
-  readonly resume: (sessionID: string) => Effect.Effect<void, unknown>
-  readonly wait: (sessionID: string) => Effect.Effect<void, unknown>
-  readonly interrupt: (sessionID: string) => Effect.Effect<void, unknown>
-}

packages/plugin/src/v2/promise/context.ts

@@ -6,7 +6,6 @@ import type { CommandHooks } from "./command.js"
 import type { IntegrationHooks } from "./integration.js"
 import type { PluginDomain } from "./plugin.js"
 import type { ReferenceHooks } from "./reference.js"
-import type { SessionDomain } from "./session.js"
 import type { SkillHooks } from "./skill.js"
 import type { Reload } from "./registration.js"
 
@@ -19,6 +18,5 @@ export interface PluginContext {
   readonly integration: IntegrationHooks & Reload
   readonly plugin: PluginDomain
   readonly reference: ReferenceHooks & Reload
-  readonly session: SessionDomain
   readonly skill: SkillHooks & Reload
 }

packages/plugin/src/v2/promise/session.ts

@@ -1,29 +0,0 @@
-import type { PromptInput, SessionInputAdmitted, SessionMessage, SessionV2Info } from "@opencode-ai/sdk/v2/types"
-
-export interface SessionDomain {
-  readonly create: (input: {
-    readonly id?: string
-    readonly parentID?: string
-    readonly title?: string
-    readonly agent?: string
-    readonly model?: SessionV2Info["model"]
-  }) => Promise<SessionV2Info>
-  readonly get: (sessionID: string) => Promise<SessionV2Info>
-  readonly messages: (input: {
-    readonly sessionID: string
-    readonly limit?: number
-    readonly order?: "asc" | "desc"
-    readonly cursor?: { readonly id: string; readonly direction: "previous" | "next" }
-  }) => Promise<ReadonlyArray<SessionMessage>>
-  readonly context: (sessionID: string) => Promise<ReadonlyArray<SessionMessage>>
-  readonly prompt: (input: {
-    readonly id?: string
-    readonly sessionID: string
-    readonly prompt: PromptInput
-    readonly delivery?: "steer" | "queue"
-    readonly resume?: boolean
-  }) => Promise<SessionInputAdmitted>
-  readonly resume: (sessionID: string) => Promise<void>
-  readonly wait: (sessionID: string) => Promise<void>
-  readonly interrupt: (sessionID: string) => Promise<void>
-}

packages/protocol/src/api.ts

@@ -1,6 +1,7 @@
 import { Context } from "effect"
 import { HttpApi, HttpApiGroup, HttpApiMiddleware, OpenApi } from "effect/unstable/httpapi"
 import { SchemaErrorMiddleware } from "./middleware/schema-error"
+import { GenerateGroup } from "./groups/generate"
 import { MessageGroup } from "./groups/message"
 import { ModelGroup } from "./groups/model"
 import { ProviderGroup } from "./groups/provider"
@@ -14,6 +15,7 @@ import type { Definition } from "@opencode-ai/schema/event"
 import { AgentGroup } from "./groups/agent"
 import { HealthGroup } from "./groups/health"
 import { PtyGroup } from "./groups/pty"
+import { ShellGroup } from "./groups/shell"
 import { makeQuestionGroup } from "./groups/question"
 import { ReferenceGroup } from "./groups/reference"
 import { Authorization } from "./middleware/authorization"
@@ -41,6 +43,7 @@ const makeApiFromGroup = <
     .add(makeSessionGroup(sessionLocationMiddleware))
     .add(MessageGroup.middleware(sessionLocationMiddleware))
     .add(ModelGroup.middleware(locationMiddleware))
+    .add(GenerateGroup.middleware(locationMiddleware))
     .add(ProviderGroup.middleware(locationMiddleware))
     .add(IntegrationGroup.middleware(locationMiddleware))
     .add(CredentialGroup.middleware(locationMiddleware))
@@ -50,6 +53,7 @@ const makeApiFromGroup = <
     .add(SkillGroup.middleware(locationMiddleware))
     .add(eventGroup)
     .add(PtyGroup.middleware(locationMiddleware))
+    .add(ShellGroup.middleware(locationMiddleware))
     .add(makeQuestionGroup(locationMiddleware, sessionLocationMiddleware))
     .add(ReferenceGroup.middleware(locationMiddleware))
     .add(ProjectCopyGroup.middleware(locationMiddleware))

packages/protocol/src/errors.ts

@@ -118,3 +118,12 @@ export class PtyNotFoundError extends Schema.TaggedErrorClass<PtyNotFoundError>(
   },
   { httpApiStatus: 404 },
 ) {}
+
+export class ShellNotFoundError extends Schema.TaggedErrorClass<ShellNotFoundError>()(
+  "ShellNotFoundError",
+  {
+    id: Schema.String,
+    message: Schema.String,
+  },
+  { httpApiStatus: 404 },
+) {}

packages/protocol/src/groups/generate.ts

@@ -0,0 +1,35 @@
+import { Model } from "@opencode-ai/schema/model"
+import { Schema } from "effect"
+import { HttpApiEndpoint, HttpApiGroup, OpenApi } from "effect/unstable/httpapi"
+import { InvalidRequestError, ServiceUnavailableError } from "../errors"
+import { LocationQuery, locationQueryOpenApi } from "./location"
+
+export const GenerateGroup = HttpApiGroup.make("server.generate")
+  .add(
+    HttpApiEndpoint.post("generate.text", "/api/generate", {
+      query: LocationQuery,
+      payload: Schema.Struct({
+        prompt: Schema.String,
+        model: Model.Ref.pipe(Schema.optional),
+      }),
+      success: Schema.Struct({
+        data: Schema.Struct({ text: Schema.String }),
+      }).annotate({ identifier: "GenerateTextResponse" }),
+      error: [InvalidRequestError, ServiceUnavailableError],
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.generate.text",
+          summary: "Generate text",
+          description:
+            "Run one stateless model generation at the requested location and return the assistant text. Uses the location's default model when none is specified.",
+        }),
+      ),
+  )
+  .annotateMerge(
+    OpenApi.annotations({
+      title: "generate",
+      description: "Experimental one-shot generation routes.",
+    }),
+  )

packages/protocol/src/groups/session.ts

@@ -243,7 +243,7 @@ export const makeSessionGroup = <I extends HttpApiMiddleware.AnyId, S>(sessionLo
       HttpApiEndpoint.post("session.compact", "/api/session/:sessionID/compact", {
         params: { sessionID: Session.ID },
         success: HttpApiSchema.NoContent,
-        error: [SessionNotFoundError, ServiceUnavailableError],
+        error: [SessionNotFoundError, SessionBusyError, ServiceUnavailableError, UnknownError],
       })
         .middleware(sessionLocationMiddleware)
         .annotateMerge(

packages/protocol/src/groups/shell.ts

@@ -0,0 +1,89 @@
+import { Shell } from "@opencode-ai/schema/shell"
+import { Location } from "@opencode-ai/schema/location"
+import { Schema } from "effect"
+import { HttpApiEndpoint, HttpApiGroup, HttpApiSchema, OpenApi } from "effect/unstable/httpapi"
+import { ShellNotFoundError } from "../errors"
+import { LocationQuery, locationQueryOpenApi } from "./location"
+
+export const ShellGroup = HttpApiGroup.make("server.shell")
+  .add(
+    HttpApiEndpoint.get("shell.list", "/api/shell", {
+      query: LocationQuery,
+      success: Location.response(Schema.Array(Shell.Info)),
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.shell.list",
+          summary: "List running shell commands",
+          description: "List currently running shell commands for a location. Exited commands are not included.",
+        }),
+      ),
+  )
+  .add(
+    HttpApiEndpoint.post("shell.create", "/api/shell", {
+      query: LocationQuery,
+      payload: Shell.CreateInput,
+      success: Location.response(Shell.Info),
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.shell.create",
+          summary: "Run shell command",
+          description:
+            "Spawn one non-interactive shell command for a location. Combined stdout/stderr is captured to a file pageable via output.",
+        }),
+      ),
+  )
+  .add(
+    HttpApiEndpoint.get("shell.get", "/api/shell/:id", {
+      params: { id: Shell.ID },
+      query: LocationQuery,
+      success: Location.response(Shell.Info),
+      error: ShellNotFoundError,
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.shell.get",
+          summary: "Get shell command",
+          description: "Get one shell command, including its status and exit code once exited.",
+        }),
+      ),
+  )
+  .add(
+    HttpApiEndpoint.get("shell.output", "/api/shell/:id/output", {
+      params: { id: Shell.ID },
+      query: Schema.Struct({ ...LocationQuery.fields, ...Shell.OutputInput.fields }),
+      success: Location.response(Shell.Output),
+      error: ShellNotFoundError,
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.shell.output",
+          summary: "Read shell output",
+          description: "Page through captured combined output by absolute byte cursor.",
+        }),
+      ),
+  )
+  .add(
+    HttpApiEndpoint.delete("shell.remove", "/api/shell/:id", {
+      params: { id: Shell.ID },
+      query: LocationQuery,
+      success: HttpApiSchema.NoContent,
+      error: ShellNotFoundError,
+    })
+      .annotateMerge(locationQueryOpenApi)
+      .annotateMerge(
+        OpenApi.annotations({
+          identifier: "v2.shell.remove",
+          summary: "Remove shell command",
+          description: "Terminate and remove one shell command and its retained output.",
+        }),
+      ),
+  )
+  .annotateMerge(
+    OpenApi.annotations({ title: "shell", description: "Experimental location-scoped shell command routes." }),
+  )

packages/schema/src/event-manifest.ts

@@ -21,6 +21,7 @@ import { Question } from "./question"
 import { QuestionV1 } from "./question-v1"
 import { Reference } from "./reference"
 import { ServerEvent } from "./server-event"
+import { Shell } from "./shell"
 import { SessionCompactionEvent } from "./session-compaction-event"
 import { SessionEvent } from "./session-event"
 import { SessionStatusEvent } from "./session-status-event"
@@ -51,6 +52,7 @@ const featureDefinitions = Event.inventory(
   ...ProjectDirectories.Event.Definitions,
   ...FileSystemWatcher.Event.Definitions,
   ...Pty.Event.Definitions,
+  ...Shell.Event.Definitions,
   ...Question.Event.Definitions,
 )
 

packages/schema/src/index.ts

@@ -18,6 +18,7 @@ export { Revert } from "./revert"
 export { Session } from "./session"
 export { SessionInput } from "./session-input"
 export { SessionMessage } from "./session-message"
+export { Shell } from "./shell"
 export { Skill } from "./skill"
 export { Pty } from "./pty"
 export { PtyTicket } from "./pty-ticket"

packages/schema/src/prompt-input.ts

@@ -23,5 +23,4 @@ export const Prompt = Schema.Struct({
   text: Schema.String,
   files: Schema.Array(FileAttachment).pipe(optional),
   agents: Schema.Array(AgentAttachment).pipe(optional),
-  system: Schema.String.pipe(optional),
 }).annotate({ identifier: "PromptInput" })

packages/schema/src/prompt.ts

@@ -42,18 +42,16 @@ export const Prompt = Schema.Struct({
   text: Schema.String,
   files: Schema.Array(FileAttachment).pipe(optional),
   agents: Schema.Array(AgentAttachment).pipe(optional),
-  system: Schema.String.pipe(optional),
 })
   .annotate({ identifier: "Prompt" })
   .pipe(
     statics((schema) => ({
       equivalence: Schema.toEquivalence(schema),
-      fromUserMessage: (input: Pick<Prompt, "text" | "files" | "agents" | "system">) =>
+      fromUserMessage: (input: Pick<Prompt, "text" | "files" | "agents">) =>
         schema.make({
           text: input.text,
           ...(input.files === undefined ? {} : { files: input.files }),
           ...(input.agents === undefined ? {} : { agents: input.agents }),
-          ...(input.system === undefined ? {} : { system: input.system }),
         }),
     })),
   )

packages/schema/src/session-message.ts

@@ -47,7 +47,6 @@ export const User = Schema.Struct({
   text: Prompt.fields.text,
   files: Prompt.fields.files,
   agents: Prompt.fields.agents,
-  system: Prompt.fields.system,
   type: Schema.Literal("user"),
 }).annotate({ identifier: "Session.Message.User" })
 

packages/schema/src/shell.ts

@@ -0,0 +1,79 @@
+export * as Shell from "./shell"
+
+import { Schema } from "effect"
+import { optional } from "./schema"
+import { define, inventory } from "./event"
+import { ascending } from "./identifier"
+import { NonNegativeInt, statics } from "./schema"
+
+const IDSchema = Schema.String.check(Schema.isStartsWith("sh_")).pipe(Schema.brand("ShellID"))
+
+export const ID = IDSchema.pipe(
+  statics((schema: typeof IDSchema) => {
+    const create = () => schema.make("sh_" + ascending())
+    return {
+      create,
+      ascending: (id?: string) => (id === undefined ? create() : schema.make(id)),
+    }
+  }),
+)
+export type ID = typeof ID.Type
+
+export const Status = Schema.Literals(["running", "exited", "timeout", "killed"])
+export type Status = typeof Status.Type
+
+export const Time = Schema.Struct({
+  started: Schema.Number,
+  completed: optional(Schema.Number),
+})
+export interface Time extends Schema.Schema.Type<typeof Time> {}
+
+// Opaque caller-supplied tags echoed back on Info and events. The Shell service never interprets
+// these; callers (e.g. ShellTool stores the originating session ID) use them to filter or correlate.
+export const Metadata = Schema.Record(Schema.String, Schema.Unknown)
+export type Metadata = typeof Metadata.Type
+
+export const Info = Schema.Struct({
+  id: ID,
+  status: Status,
+  command: Schema.String,
+  cwd: Schema.String,
+  shell: Schema.String,
+  // Absolute path of the file capturing combined stdout/stderr. Page through it via `output`.
+  file: Schema.String,
+  pid: optional(NonNegativeInt),
+  exit: optional(Schema.Number),
+  // Always present; defaults to an empty object when the creator supplies no metadata.
+  metadata: Metadata,
+  time: Time,
+}).annotate({ identifier: "Shell" })
+export interface Info extends Schema.Schema.Type<typeof Info> {}
+
+const Created = define({ type: "shell.created", schema: { info: Info } })
+const Exited = define({ type: "shell.exited", schema: { id: ID, exit: optional(Schema.Number), status: Status } })
+const Deleted = define({ type: "shell.deleted", schema: { id: ID } })
+export const Event = { Created, Exited, Deleted, Definitions: inventory(Created, Exited, Deleted) }
+
+export const CreateInput = Schema.Struct({
+  command: Schema.String,
+  cwd: optional(Schema.String),
+  timeout: optional(NonNegativeInt),
+  metadata: optional(Metadata),
+})
+export interface CreateInput extends Schema.Schema.Type<typeof CreateInput> {}
+
+export const OutputInput = Schema.Struct({
+  cursor: optional(NonNegativeInt),
+  limit: optional(NonNegativeInt),
+})
+export interface OutputInput extends Schema.Schema.Type<typeof OutputInput> {}
+
+export const Output = Schema.Struct({
+  output: Schema.String,
+  // Absolute cursor after this page. Equals `size` once fully caught up.
+  cursor: NonNegativeInt,
+  // Total bytes captured so far. A consumer has more to page while `cursor < size`.
+  size: NonNegativeInt,
+  truncated: Schema.Boolean,
+})
+export interface Output extends Schema.Schema.Type<typeof Output> {}

packages/sdk-next/src/opencode.ts

@@ -1,5 +1,6 @@
 import { OpenCode } from "@opencode-ai/client/effect"
 import { PermissionSaved } from "@opencode-ai/core/permission/saved"
+import { SdkPlugins } from "@opencode-ai/core/plugin/sdk"
 import { ApplicationTools } from "@opencode-ai/core/tool/application-tools"
 import { createEmbeddedRoutes } from "@opencode-ai/server/routes"
 import { Context, Effect, Layer, Scope } from "effect"
@@ -9,11 +10,12 @@ export const create = Effect.fn("OpenCode.create")(function* () {
   const scope = yield* Scope.Scope
   const memoMap = yield* Layer.makeMemoMap
   const context = yield* Layer.buildWithMemoMap(
-    Layer.merge(ApplicationTools.layer, PermissionSaved.defaultLayer),
+    Layer.mergeAll(ApplicationTools.layer, PermissionSaved.defaultLayer, SdkPlugins.layer),
     memoMap,
     scope,
   )
   const tools = Context.get(context, ApplicationTools.Service)
+  const plugins = Context.get(context, SdkPlugins.Service)
   const permissions = Context.get(context, PermissionSaved.Service)
   const web = yield* Effect.acquireRelease(
     Effect.sync(() =>
@@ -37,6 +39,12 @@ export const create = Effect.fn("OpenCode.create")(function* () {
   return {
     ...client,
     tools: { register: tools.register },
+    // The embedded host contributes plugins through the ordinary discovery flow:
+    // each plugin's `effect` runs inside every Location with the real
+    // `PluginContext`, so `ctx.agent.transform` and every other hook behave exactly
+    // as they do for a config-discovered plugin. Define agent profiles here at
+    // startup, then select one per Session with `sessions.create({ agent })`.
+    plugin: plugins.register,
   }
 })
 

packages/sdk/js/src/v2/gen/sdk.gen.ts

@@ -385,6 +385,16 @@ import type {
   V2SessionSwitchModelResponses,
   V2SessionWaitErrors,
   V2SessionWaitResponses,
+  V2ShellCreateErrors,
+  V2ShellCreateResponses,
+  V2ShellGetErrors,
+  V2ShellGetResponses,
+  V2ShellListErrors,
+  V2ShellListResponses,
+  V2ShellOutputErrors,
+  V2ShellOutputResponses,
+  V2ShellRemoveErrors,
+  V2ShellRemoveResponses,
   V2SkillListErrors,
   V2SkillListResponses,
   VcsApplyErrors,
@@ -6849,6 +6859,179 @@ export class Pty2 extends HeyApiClient {
   }
 }
 
+export class Shell extends HeyApiClient {
+  /**
+   * List running shell commands
+   *
+   * List currently running shell commands for a location. Exited commands are not included.
+   */
+  public list<ThrowOnError extends boolean = false>(
+    parameters?: {
+      location?: {
+        directory?: string
+        workspace?: string
+      }
+    },
+    options?: Options<never, ThrowOnError>,
+  ) {
+    const params = buildClientParams([parameters], [{ args: [{ in: "query", key: "location" }] }])
+    return (options?.client ?? this.client).get<V2ShellListResponses, V2ShellListErrors, ThrowOnError>({
+      url: "/api/shell",
+      ...options,
+      ...params,
+    })
+  }
+
+  /**
+   * Run shell command
+   *
+   * Spawn one non-interactive shell command for a location. Combined stdout/stderr is captured to a file pageable via output.
+   */
+  public create<ThrowOnError extends boolean = false>(
+    parameters?: {
+      location?: {
+        directory?: string
+        workspace?: string
+      }
+      command?: string
+      cwd?: string
+      timeout?: number
+      metadata?: {
+        [key: string]: unknown
+      }
+    },
+    options?: Options<never, ThrowOnError>,
+  ) {
+    const params = buildClientParams(
+      [parameters],
+      [
+        {
+          args: [
+            { in: "query", key: "location" },
+            { in: "body", key: "command" },
+            { in: "body", key: "cwd" },
+            { in: "body", key: "timeout" },
+            { in: "body", key: "metadata" },
+          ],
+        },
+      ],
+    )
+    return (options?.client ?? this.client).post<V2ShellCreateResponses, V2ShellCreateErrors, ThrowOnError>({
+      url: "/api/shell",
+      ...options,
+      ...params,
+      headers: {
+        "Content-Type": "application/json",
+        ...options?.headers,
+        ...params.headers,
+      },
+    })
+  }
+
+  /**
+   * Remove shell command
+   *
+   * Terminate and remove one shell command and its retained output.
+   */
+  public remove<ThrowOnError extends boolean = false>(
+    parameters: {
+      id: string
+      location?: {
+        directory?: string
+        workspace?: string
+      }
+    },
+    options?: Options<never, ThrowOnError>,
+  ) {
+    const params = buildClientParams(
+      [parameters],
+      [
+        {
+          args: [
+            { in: "path", key: "id" },
+            { in: "query", key: "location" },
+          ],
+        },
+      ],
+    )
+    return (options?.client ?? this.client).delete<V2ShellRemoveResponses, V2ShellRemoveErrors, ThrowOnError>({
+      url: "/api/shell/{id}",
+      ...options,
+      ...params,
+    })
+  }
+
+  /**
+   * Get shell command
+   *
+   * Get one shell command, including its status and exit code once exited.
+   */
+  public get<ThrowOnError extends boolean = false>(
+    parameters: {
+      id: string
+      location?: {
+        directory?: string
+        workspace?: string
+      }
+    },
+    options?: Options<never, ThrowOnError>,
+  ) {
+    const params = buildClientParams(
+      [parameters],
+      [
+        {
+          args: [
+            { in: "path", key: "id" },
+            { in: "query", key: "location" },
+          ],
+        },
+      ],
+    )
+    return (options?.client ?? this.client).get<V2ShellGetResponses, V2ShellGetErrors, ThrowOnError>({
+      url: "/api/shell/{id}",
+      ...options,
+      ...params,
+    })
+  }
+
+  /**
+   * Read shell output
+   *
+   * Page through captured combined output by absolute byte cursor.
+   */
+  public output<ThrowOnError extends boolean = false>(
+    parameters: {
+      id: string
+      location?: {
+        directory?: string
+        workspace?: string
+      }
+      cursor?: string
+      limit?: string
+    },
+    options?: Options<never, ThrowOnError>,
+  ) {
+    const params = buildClientParams(
+      [parameters],
+      [
+        {
+          args: [
+            { in: "path", key: "id" },
+            { in: "query", key: "location" },
+            { in: "query", key: "cursor" },
+            { in: "query", key: "limit" },
+          ],
+        },
+      ],
+    )
+    return (options?.client ?? this.client).get<V2ShellOutputResponses, V2ShellOutputErrors, ThrowOnError>({
+      url: "/api/shell/{id}/output",
+      ...options,
+      ...params,
+    })
+  }
+}
+
 export class Request2 extends HeyApiClient {
   /**
    * List pending question requests
@@ -7095,6 +7278,11 @@ export class V2 extends HeyApiClient {
     return (this._pty ??= new Pty2({ client: this.client }))
   }
 
+  private _shell?: Shell
+  get shell(): Shell {
+    return (this._shell ??= new Shell({ client: this.client }))
+  }
+
   private _question?: Question3
   get question(): Question3 {
     return (this._question ??= new Question3({ client: this.client }))

packages/sdk/js/src/v2/gen/types.gen.ts

@@ -65,6 +65,9 @@ export type Event =
   | EventPtyUpdated
   | EventPtyExited
   | EventPtyDeleted
+  | EventShellCreated
+  | EventShellExited
+  | EventShellDeleted
   | EventQuestionV2Asked
   | EventQuestionV2Replied
   | EventQuestionV2Rejected
@@ -643,6 +646,7 @@ export type Prompt = {
   text: string
   files?: Array<PromptFileAttachment>
   agents?: Array<PromptAgentAttachment>
+  system?: string
 }
 
 export type Pty = {
@@ -656,6 +660,24 @@ export type Pty = {
   exitCode?: number
 }
 
+export type Shell = {
+  id: string
+  status: "running" | "exited" | "timeout" | "killed"
+  command: string
+  cwd: string
+  shell: string
+  file: string
+  pid?: number
+  exit?: number | "NaN" | "Infinity" | "-Infinity" | "Infinity" | "-Infinity" | "NaN"
+  metadata: {
+    [key: string]: unknown
+  }
+  time: {
+    started: number | "NaN" | "Infinity" | "-Infinity" | "Infinity" | "-Infinity" | "NaN"
+    completed?: number | "NaN" | "Infinity" | "-Infinity" | "Infinity" | "-Infinity" | "NaN"
+  }
+}
+
 export type Todo = {
   /**
    * Brief description of the task
@@ -1338,6 +1360,29 @@ export type GlobalEvent = {
           id: string
         }
       }
+    | {
+        id: string
+        type: "shell.created"
+        properties: {
+          info: Shell
+        }
+      }
+    | {
+        id: string
+        type: "shell.exited"
+        properties: {
+          id: string
+          exit?: number | "NaN" | "Infinity" | "-Infinity" | "Infinity" | "-Infinity" | "NaN"
+          status: "running" | "exited" | "timeout" | "killed"
+        }
+      }
+    | {
+        id: string
+        type: "shell.deleted"
+        properties: {
+          id: string
+        }
+      }
     | {
         id: string
         type: "question.v2.asked"
@@ -2715,6 +2760,7 @@ export type PromptInput = {
   text: string
   files?: Array<PromptInputFileAttachment>
   agents?: Array<PromptAgentAttachment>
+  system?: string
 }
 
 export type ConflictError = {
@@ -2729,6 +2775,12 @@ export type ServiceUnavailableError = {
   service?: string
 }
 
+export type UnknownError1 = {
+  _tag: "UnknownError"
+  message: string
+  ref?: string
+}
+
 export type MessageNotFoundError = {
   _tag: "MessageNotFoundError"
   sessionID: string
@@ -2736,12 +2788,6 @@ export type MessageNotFoundError = {
   message: string
 }
 
-export type UnknownError1 = {
-  _tag: "UnknownError"
-  message: string
-  ref?: string
-}
-
 export type SessionDurableEvent =
   | SessionNextAgentSwitched
   | SessionNextModelSwitched
@@ -2804,6 +2850,24 @@ export type OutputFormat1 =
       retryCount?: number
     }
 
+export type Shell1 = {
+  id: string
+  status: "running" | "exited" | "timeout" | "killed"
+  command: string
+  cwd: string
+  shell: string
+  file: string
+  pid?: number
+  exit?: number | "NaN" | "Infinity" | "-Infinity"
+  metadata: {
+    [key: string]: unknown
+  }
+  time: {
+    started: number | "NaN" | "Infinity" | "-Infinity"
+    completed?: number | "NaN" | "Infinity" | "-Infinity"
+  }
+}
+
 export type SessionStatus2 = {
   id: string
   metadata?: {
@@ -2920,6 +2984,9 @@ export type V2Event =
   | PtyUpdated
   | PtyExited
   | PtyDeleted
+  | ShellCreated
+  | ShellExited
+  | ShellDeleted
   | QuestionV2Asked
   | QuestionV2Replied
   | QuestionV2Rejected
@@ -2957,6 +3024,12 @@ export type ForbiddenError = {
   message: string
 }
 
+export type ShellNotFoundError = {
+  _tag: "ShellNotFoundError"
+  id: string
+  message: string
+}
+
 export type ProjectCopyError = {
   name: "ProjectCopyError"
   data: {
@@ -2969,6 +3042,24 @@ export type EffectHttpApiErrorForbidden = {
   _tag: "Forbidden"
 }
 
+export type Shell2 = {
+  id: string
+  status: "running" | "exited" | "timeout" | "killed"
+  command: string
+  cwd: string
+  shell: string
+  file: string
+  pid?: number
+  exit?: number | "NaN" | "Infinity" | "-Infinity"
+  metadata: {
+    [key: string]: unknown
+  }
+  time: {
+    started: number | "NaN" | "Infinity" | "-Infinity"
+    completed?: number | "NaN" | "Infinity" | "-Infinity"
+  }
+}
+
 export type EventTuiPromptAppend2 = {
   id: string
   type: "tui.prompt.append"
@@ -4005,6 +4096,7 @@ export type SessionMessageUser = {
   text: string
   files?: Array<PromptFileAttachment>
   agents?: Array<PromptAgentAttachment>
+  system?: string
   type: "user"
 }
 
@@ -5641,6 +5733,59 @@ export type PtyDeleted = {
   }
 }
 
+export type ShellCreated = {
+  id: string
+  metadata?: {
+    [key: string]: unknown
+  }
+  type: "shell.created"
+  durable?: {
+    aggregateID: string
+    seq: number
+    version: number
+  }
+  location?: LocationRef
+  data: {
+    info: Shell1
+  }
+}
+
+export type ShellExited = {
+  id: string
+  metadata?: {
+    [key: string]: unknown
+  }
+  type: "shell.exited"
+  durable?: {
+    aggregateID: string
+    seq: number
+    version: number
+  }
+  location?: LocationRef
+  data: {
+    id: string
+    exit?: number | "NaN" | "Infinity" | "-Infinity"
+    status: "running" | "exited" | "timeout" | "killed"
+  }
+}
+
+export type ShellDeleted = {
+  id: string
+  metadata?: {
+    [key: string]: unknown
+  }
+  type: "shell.deleted"
+  durable?: {
+    aggregateID: string
+    seq: number
+    version: number
+  }
+  location?: LocationRef
+  data: {
+    id: string
+  }
+}
+
 export type QuestionV2Asked = {
   id: string
   metadata?: {
@@ -6859,6 +7004,32 @@ export type EventPtyDeleted = {
   }
 }
 
+export type EventShellCreated = {
+  id: string
+  type: "shell.created"
+  properties: {
+    info: Shell2
+  }
+}
+
+export type EventShellExited = {
+  id: string
+  type: "shell.exited"
+  properties: {
+    id: string
+    exit?: number | "NaN" | "Infinity" | "-Infinity"
+    status: "running" | "exited" | "timeout" | "killed"
+  }
+}
+
+export type EventShellDeleted = {
+  id: string
+  type: "shell.deleted"
+  properties: {
+    id: string
+  }
+}
+
 export type EventQuestionV2Asked = {
   id: string
   type: "question.v2.asked"
@@ -11708,6 +11879,14 @@ export type V2SessionCompactErrors = {
    * SessionNotFoundError
    */
   404: SessionNotFoundError
+  /**
+   * SessionBusyError
+   */
+  409: SessionBusyError
+  /**
+   * UnknownError
+   */
+  500: UnknownError1
   /**
    * ServiceUnavailableError
    */
@@ -13401,6 +13580,220 @@ export type V2PtyConnectResponses = {
 
 export type V2PtyConnectResponse = V2PtyConnectResponses[keyof V2PtyConnectResponses]
 
+export type V2ShellListData = {
+  body?: never
+  path?: never
+  query?: {
+    location?: {
+      directory?: string
+      workspace?: string
+    }
+  }
+  url: "/api/shell"
+}
+
+export type V2ShellListErrors = {
+  /**
+   * InvalidRequestError
+   */
+  400: InvalidRequestError
+  /**
+   * UnauthorizedError
+   */
+  401: UnauthorizedError
+}
+
+export type V2ShellListError = V2ShellListErrors[keyof V2ShellListErrors]
+
+export type V2ShellListResponses = {
+  /**
+   * Success
+   */
+  200: {
+    location: LocationInfo
+    data: Array<Shell>
+  }
+}
+
+export type V2ShellListResponse = V2ShellListResponses[keyof V2ShellListResponses]
+
+export type V2ShellCreateData = {
+  body: {
+    command: string
+    cwd?: string
+    timeout?: number
+    metadata?: {
+      [key: string]: unknown
+    }
+  }
+  path?: never
+  query?: {
+    location?: {
+      directory?: string
+      workspace?: string
+    }
+  }
+  url: "/api/shell"
+}
+
+export type V2ShellCreateErrors = {
+  /**
+   * InvalidRequestError
+   */
+  400: InvalidRequestError
+  /**
+   * UnauthorizedError
+   */
+  401: UnauthorizedError
+}
+
+export type V2ShellCreateError = V2ShellCreateErrors[keyof V2ShellCreateErrors]
+
+export type V2ShellCreateResponses = {
+  /**
+   * Success
+   */
+  200: {
+    location: LocationInfo
+    data: Shell
+  }
+}
+
+export type V2ShellCreateResponse = V2ShellCreateResponses[keyof V2ShellCreateResponses]
+
+export type V2ShellRemoveData = {
+  body?: never
+  path: {
+    id: string
+  }
+  query?: {
+    location?: {
+      directory?: string
+      workspace?: string
+    }
+  }
+  url: "/api/shell/{id}"
+}
+
+export type V2ShellRemoveErrors = {
+  /**
+   * InvalidRequestError
+   */
+  400: InvalidRequestError
+  /**
+   * UnauthorizedError
+   */
+  401: UnauthorizedError
+  /**
+   * ShellNotFoundError
+   */
+  404: ShellNotFoundError
+}
+
+export type V2ShellRemoveError = V2ShellRemoveErrors[keyof V2ShellRemoveErrors]
+
+export type V2ShellRemoveResponses = {
+  /**
+   * <No Content>
+   */
+  204: void
+}
+
+export type V2ShellRemoveResponse = V2ShellRemoveResponses[keyof V2ShellRemoveResponses]
+
+export type V2ShellGetData = {
+  body?: never
+  path: {
+    id: string
+  }
+  query?: {
+    location?: {
+      directory?: string
+      workspace?: string
+    }
+  }
+  url: "/api/shell/{id}"
+}
+
+export type V2ShellGetErrors = {
+  /**
+   * InvalidRequestError
+   */
+  400: InvalidRequestError
+  /**
+   * UnauthorizedError
+   */
+  401: UnauthorizedError
+  /**
+   * ShellNotFoundError
+   */
+  404: ShellNotFoundError
+}
+
+export type V2ShellGetError = V2ShellGetErrors[keyof V2ShellGetErrors]
+
+export type V2ShellGetResponses = {
+  /**
+   * Success
+   */
+  200: {
+    location: LocationInfo
+    data: Shell
+  }
+}
+
+export type V2ShellGetResponse = V2ShellGetResponses[keyof V2ShellGetResponses]
+
+export type V2ShellOutputData = {
+  body?: never
+  path: {
+    id: string
+  }
+  query?: {
+    location?: {
+      directory?: string
+      workspace?: string
+    }
+    cursor?: string
+    limit?: string
+  }
+  url: "/api/shell/{id}/output"
+}
+
+export type V2ShellOutputErrors = {
+  /**
+   * InvalidRequestError
+   */
+  400: InvalidRequestError
+  /**
+   * UnauthorizedError
+   */
+  401: UnauthorizedError
+  /**
+   * ShellNotFoundError
+   */
+  404: ShellNotFoundError
+}
+
+export type V2ShellOutputError = V2ShellOutputErrors[keyof V2ShellOutputErrors]
+
+export type V2ShellOutputResponses = {
+  /**
+   * Success
+   */
+  200: {
+    location: LocationInfo
+    data: {
+      output: string
+      cursor: number
+      size: number
+      truncated: boolean
+    }
+  }
+}
+
+export type V2ShellOutputResponse = V2ShellOutputResponses[keyof V2ShellOutputResponses]
+
 export type V2QuestionRequestListData = {
   body?: never
   path?: never

packages/server/src/handlers.ts

@@ -1,4 +1,5 @@
 import { Layer } from "effect"
+import { GenerateHandler } from "./handlers/generate"
 import { MessageHandler } from "./handlers/message"
 import { ModelHandler } from "./handlers/model"
 import { ProviderHandler } from "./handlers/provider"
@@ -11,6 +12,7 @@ import { EventHandler } from "./handlers/event"
 import { AgentHandler } from "./handlers/agent"
 import { HealthHandler } from "./handlers/health"
 import { PtyHandler } from "./handlers/pty"
+import { ShellHandler } from "./handlers/shell"
 import { QuestionHandler } from "./handlers/question"
 import { ReferenceHandler } from "./handlers/reference"
 import { LocationHandler } from "./handlers/location"
@@ -25,6 +27,7 @@ export const handlers = Layer.mergeAll(
   SessionHandler,
   MessageHandler,
   ModelHandler,
+  GenerateHandler,
   ProviderHandler,
   IntegrationHandler,
   CredentialHandler,
@@ -34,6 +37,7 @@ export const handlers = Layer.mergeAll(
   SkillHandler,
   EventHandler,
   PtyHandler,
+  ShellHandler,
   QuestionHandler,
   ReferenceHandler,
   ProjectCopyHandler,

packages/server/src/handlers/generate.ts

@@ -0,0 +1,26 @@
+import { Generate } from "@opencode-ai/core/generate"
+import { InvalidRequestError, ServiceUnavailableError } from "@opencode-ai/protocol/errors"
+import { Effect } from "effect"
+import { HttpApiBuilder } from "effect/unstable/httpapi"
+import { Api } from "../api"
+
+export const GenerateHandler = HttpApiBuilder.group(Api, "server.generate", (handlers) =>
+  Effect.gen(function* () {
+    return handlers.handle(
+      "generate.text",
+      Effect.fn("server.generate.text")(function* (request) {
+        const generate = yield* Generate.Service
+        const text = yield* generate
+          .text(request.payload)
+          .pipe(
+            Effect.mapError((error) =>
+              error._tag === "Generate.ModelSelectionError"
+                ? new InvalidRequestError({ message: error.message })
+                : new ServiceUnavailableError({ message: error.message, service: error.service }),
+            ),
+          )
+        return { data: { text } }
+      }),
+    )
+  }),
+)

packages/server/src/handlers/session.ts

@@ -1,6 +1,4 @@
 import { SessionV2 } from "@opencode-ai/core/session"
-import { LocationServiceMap } from "@opencode-ai/core/location-service-map"
-import { SessionRuntime } from "@opencode-ai/core/session/runtime"
 import { DateTime, Effect, Stream } from "effect"
 import { HttpApiBuilder, HttpApiSchema } from "effect/unstable/httpapi"
 import { Api } from "../api"
@@ -22,14 +20,6 @@ const DefaultSessionHistoryLimit = 50
 export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handlers) =>
   Effect.gen(function* () {
     const session = yield* SessionV2.Service
-    const locations = yield* LocationServiceMap.Service
-    const route = Effect.fn("SessionHandler.route")(function* <A, E>(
-      sessionID: SessionV2.ID,
-      effect: Effect.Effect<A, E, SessionRuntime.Service>,
-    ) {
-      const info = yield* session.get(sessionID)
-      return yield* effect.pipe(Effect.provide(locations.get(info.location)))
-    })
 
     return handlers
       .handle(
@@ -169,18 +159,15 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
         "session.prompt",
         Effect.fn(function* (ctx) {
           return {
-            data: yield* route(
-              ctx.params.sessionID,
-              SessionRuntime.Service.use((runtime) =>
-                runtime.prompt({
-                  sessionID: ctx.params.sessionID,
-                  id: ctx.payload.id,
-                  prompt: ctx.payload.prompt,
-                  delivery: ctx.payload.delivery,
-                  resume: ctx.payload.resume,
-                }),
-              ),
-            ).pipe(
+            data: yield* session
+              .prompt({
+                sessionID: ctx.params.sessionID,
+                id: ctx.payload.id,
+                prompt: ctx.payload.prompt,
+                delivery: ctx.payload.delivery,
+                resume: ctx.payload.resume,
+              })
+              .pipe(
                 Effect.catchTag("Session.NotFoundError", (error) =>
                   Effect.fail(
                     new SessionNotFoundError({
@@ -221,6 +208,25 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
                 }),
               ),
             ),
+            Effect.catchTag(
+              "Session.BusyError",
+              (error) =>
+                new SessionBusyError({
+                  sessionID: error.sessionID,
+                  message: `Session is busy: ${error.sessionID}`,
+                }),
+            ),
+            Effect.catchTag("Session.MessageDecodeError", (error) => {
+              const ref = `err_${crypto.randomUUID().slice(0, 8)}`
+              return Effect.logError("failed to decode session message during compaction").pipe(
+                Effect.annotateLogs({ ref, sessionID: error.sessionID, messageID: error.messageID }),
+                Effect.andThen(
+                  Effect.fail(
+                    new UnknownError({ message: "Unexpected server error. Check server logs for details.", ref }),
+                  ),
+                ),
+              )
+            }),
           )
           return HttpApiSchema.NoContent.make()
         }),
@@ -228,10 +234,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
       .handle(
         "session.wait",
         Effect.fn(function* (ctx) {
-          yield* route(
-            ctx.params.sessionID,
-            SessionRuntime.Service.use((runtime) => runtime.wait(ctx.params.sessionID)),
-          ).pipe(
+          yield* session.wait(ctx.params.sessionID).pipe(
             Effect.catchTag("Session.NotFoundError", (error) =>
               Effect.fail(
                 new SessionNotFoundError({
@@ -253,10 +256,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
             files: ctx.payload.files,
           })
           return {
-            data: yield* route(
-              ctx.params.sessionID,
-              SessionRuntime.Service.use((runtime) => runtime.revert.stage({ ...ctx.params, ...ctx.payload })),
-            ).pipe(
+            data: yield* session.revert.stage({ ...ctx.params, ...ctx.payload }).pipe(
               Effect.catchTag(
                 "Session.NotFoundError",
                 (error) =>
@@ -303,10 +303,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
         "session.revert.clear",
         Effect.fn(function* (ctx) {
           yield* Effect.log("session.revert.clear", { sessionID: ctx.params.sessionID })
-          yield* route(
-            ctx.params.sessionID,
-            SessionRuntime.Service.use((runtime) => runtime.revert.clear(ctx.params.sessionID)),
-          ).pipe(
+          yield* session.revert.clear(ctx.params.sessionID).pipe(
             Effect.catchTag(
               "Session.NotFoundError",
               (error) =>
@@ -344,10 +341,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
         "session.revert.commit",
         Effect.fn(function* (ctx) {
           yield* Effect.log("session.revert.commit", { sessionID: ctx.params.sessionID })
-          yield* route(
-            ctx.params.sessionID,
-            SessionRuntime.Service.use((runtime) => runtime.revert.commit(ctx.params.sessionID)),
-          ).pipe(
+          yield* session.revert.commit(ctx.params.sessionID).pipe(
             Effect.catchTag(
               "Session.NotFoundError",
               (error) =>
@@ -432,10 +426,7 @@ export const SessionHandler = HttpApiBuilder.group(Api, "server.session", (handl
       .handle(
         "session.interrupt",
         Effect.fn(function* (ctx) {
-          yield* route(
-            ctx.params.sessionID,
-            SessionRuntime.Service.use((runtime) => runtime.interrupt(ctx.params.sessionID)),
-          )
+          yield* session.interrupt(ctx.params.sessionID)
           return HttpApiSchema.NoContent.make()
         }),
       )

packages/server/src/handlers/shell.ts

@@ -0,0 +1,69 @@
+import { Shell } from "@opencode-ai/core/shell"
+import { Location } from "@opencode-ai/core/location"
+import { Effect } from "effect"
+import { HttpApiBuilder, HttpApiSchema } from "effect/unstable/httpapi"
+import { ShellNotFoundError } from "@opencode-ai/protocol/errors"
+import { Api } from "../api"
+import { response } from "../location"
+
+export const ShellHandler = HttpApiBuilder.group(Api, "server.shell", (handlers) =>
+  Effect.gen(function* () {
+    return handlers
+      .handle(
+        "shell.list",
+        Effect.fn(function* () {
+          const shell = yield* Shell.Service
+          return yield* response(shell.list())
+        }),
+      )
+      .handle(
+        "shell.create",
+        Effect.fn(function* (ctx) {
+          const shell = yield* Shell.Service
+          const location = yield* Location.Service
+          return yield* response(shell.create({ ...ctx.payload, cwd: ctx.payload.cwd || location.directory }))
+        }),
+      )
+      .handle(
+        "shell.get",
+        Effect.fn(function* (ctx) {
+          const shell = yield* Shell.Service
+          return yield* response(
+            shell.get(ctx.params.id).pipe(
+              Effect.catchTag(
+                "Shell.NotFoundError",
+                () => new ShellNotFoundError({ id: ctx.params.id, message: `Shell command not found: ${ctx.params.id}` }),
+              ),
+            ),
+          )
+        }),
+      )
+      .handle(
+        "shell.output",
+        Effect.fn(function* (ctx) {
+          const shell = yield* Shell.Service
+          return yield* response(
+            shell.output(ctx.params.id, { cursor: ctx.query.cursor, limit: ctx.query.limit }).pipe(
+              Effect.catchTag(
+                "Shell.NotFoundError",
+                () => new ShellNotFoundError({ id: ctx.params.id, message: `Shell command not found: ${ctx.params.id}` }),
+              ),
+            ),
+          )
+        }),
+      )
+      .handle(
+        "shell.remove",
+        Effect.fn(function* (ctx) {
+          const shell = yield* Shell.Service
+          yield* shell.remove(ctx.params.id).pipe(
+            Effect.catchTag(
+              "Shell.NotFoundError",
+              () => new ShellNotFoundError({ id: ctx.params.id, message: `Shell command not found: ${ctx.params.id}` }),
+            ),
+          )
+          return HttpApiSchema.NoContent.make()
+        }),
+      )
+  }),
+)

packages/tui/src/component/prompt/index.tsx

@@ -164,6 +164,9 @@ export function Prompt(props: PromptProps) {
       .filter((session) => session.parentID === props.sessionID && data.session.status(session.id) === "running")
       .length,
   )
+  const runningShells = createMemo(
+    () => data.shell.list().filter((shell) => shell.metadata.sessionID === props.sessionID).length,
+  )
   const history = usePromptHistory()
   const stash = usePromptStash()
   const keymap = useOpencodeKeymap()
@@ -284,6 +287,20 @@ export function Prompt(props: PromptProps) {
     }
   })
 
+  // Far-right footer cluster: live work counts lead, then context/cost usage, all dot-joined.
+  // When empty, the cluster falls back to the hotkey hints.
+  const statusItems = createMemo(() => {
+    const agents = activeSubagents()
+    const shells = runningShells()
+    const stats = usage()
+    return [
+      agents ? `${agents} subagent${agents === 1 ? "" : "s"}` : undefined,
+      shells ? `${shells} shell${shells === 1 ? "" : "s"}` : undefined,
+      stats?.context,
+      stats?.cost,
+    ].filter(Boolean)
+  })
+
   const [store, setStore] = createStore<{
     prompt: PromptInfo
     mode: "normal" | "shell"
@@ -1548,13 +1565,6 @@ export function Prompt(props: PromptProps) {
                     <spinner color={spinnerDef().color} frames={spinnerDef().frames} interval={40} />
                   </Show>
                 </box>
-                <Show when={activeSubagents()}>
-                  {(count) => (
-                    <Spinner color={theme.textMuted}>
-                      {count()} active subagent{count() === 1 ? "" : "s"}
-                    </Spinner>
-                  )}
-                </Show>
                 <text fg={store.interrupt > 0 ? theme.primary : theme.text}>
                   esc{" "}
                   <span style={{ fg: store.interrupt > 0 ? theme.primary : theme.textMuted }}>
@@ -1624,22 +1634,20 @@ export function Prompt(props: PromptProps) {
             <Switch>
               <Match when={store.mode === "normal"}>
                 <Switch>
-                  <Match when={usage()}>
-                    {(item) => (
-                      <text fg={theme.textMuted} wrapMode="none">
-                        {[item().context, item().cost].filter(Boolean).join(" · ")}
-                      </text>
-                    )}
+                  <Match when={statusItems().length > 0}>
+                    <text fg={theme.textMuted} wrapMode="none">
+                      {statusItems().join(" · ")}
+                    </text>
                   </Match>
                   <Match when={true}>
                     <text fg={theme.text}>
                       {agentShortcut()} <span style={{ fg: theme.textMuted }}>agents</span>
                     </text>
+                    <text fg={theme.text}>
+                      {paletteShortcut()} <span style={{ fg: theme.textMuted }}>commands</span>
+                    </text>
                   </Match>
                 </Switch>
-                <text fg={theme.text}>
-                  {paletteShortcut()} <span style={{ fg: theme.textMuted }}>commands</span>
-                </text>
               </Match>
               <Match when={store.mode === "shell"}>
                 <text fg={theme.text}>

packages/tui/src/context/data.tsx

@@ -15,6 +15,7 @@ import type {
   SessionMessageAssistantText,
   SessionMessageAssistantTool,
   SessionV2Info,
+  Shell,
   SkillV2Info,
   V2Event,
 } from "@opencode-ai/sdk/v2"
@@ -47,6 +48,9 @@ type Data = {
     permission: Record<string, PermissionSavedInfo[]>
   }
   location: Record<string, LocationData>
+  // Currently running shell commands, keyed by shell id. Entries are removed once the command
+  // exits or is deleted, so this only ever holds in-flight shells.
+  shell: Record<string, Shell>
 }
 
 function locationKey(location: LocationRef) {
@@ -72,6 +76,7 @@ export const { use: useData, provider: DataProvider } = createSimpleContext({
         permission: {},
       },
       location: {},
+      shell: {},
     })
 
     const sdk = useSDK()
@@ -467,6 +472,18 @@ export const { use: useData, provider: DataProvider } = createSimpleContext({
             ),
           )
           break
+        case "shell.created":
+          setStore("shell", event.data.info.id, event.data.info)
+          break
+        case "shell.exited":
+        case "shell.deleted":
+          setStore(
+            "shell",
+            produce((draft) => {
+              delete draft[event.data.id]
+            }),
+          )
+          break
         case "reference.updated":
           void result.location.reference.refresh()
           break
@@ -567,6 +584,23 @@ export const { use: useData, provider: DataProvider } = createSimpleContext({
           },
         },
       },
+      shell: {
+        list() {
+          return Object.values(store.shell)
+        },
+        get(id: string) {
+          return store.shell[id]
+        },
+        async refresh(ref?: LocationRef) {
+          const result = await sdk.client.v2.shell.list({ location: locationQuery(ref) }, { throwOnError: true })
+          setStore(
+            "shell",
+            produce((draft) => {
+              for (const info of result.data.data) draft[info.id] = info
+            }),
+          )
+        },
+      },
       location: {
         default() {
           return defaultLocation()
@@ -682,6 +716,7 @@ export const { use: useData, provider: DataProvider } = createSimpleContext({
         result.location.reference.refresh(),
         result.location.command.refresh(),
         result.location.skill.refresh(),
+        result.shell.refresh(),
       ])
       for (const failure of settled.filter((item) => item.status === "rejected"))
         console.error("Failed to refresh default location data", failure.reason)

packages/tui/src/feature-plugins/home/tips-view.tsx

@@ -221,8 +221,8 @@ const TIPS: Tip[] = [
   "Use {highlight}$ARGUMENTS{/highlight}, {highlight}$1{/highlight}, {highlight}$2{/highlight} in custom commands for dynamic input",
   "Use backticks in commands to inject shell output (e.g., {highlight}`git status`{/highlight})",
   "Add {highlight}.md{/highlight} files to {highlight}.opencode/agents/{/highlight} for specialized AI personas",
-  "Configure per-agent permissions for {highlight}edit{/highlight}, {highlight}bash{/highlight}, and {highlight}webfetch{/highlight} tools",
-  'Use patterns like {highlight}"git *": "allow"{/highlight} for granular bash permissions',
+  "Configure per-agent permissions for {highlight}edit{/highlight}, {highlight}shell{/highlight}, and {highlight}webfetch{/highlight} tools",
+  'Use patterns like {highlight}"git *": "allow"{/highlight} for granular shell permissions',
   'Set {highlight}"rm -rf *": "deny"{/highlight} to block destructive commands',
   'Configure {highlight}"git push": "ask"{/highlight} to require approval before pushing',
   'Set {highlight}"formatter": true{/highlight} in config to enable built-in formatters like prettier, gofmt, and ruff',
@@ -256,7 +256,7 @@ const TIPS: Tip[] = [
   "Use {highlight}instructions{/highlight} in config to load additional rules files",
   "Set agent {highlight}temperature{/highlight} from 0.0 (focused) to 1.0 (creative)",
   "Configure {highlight}steps{/highlight} to limit agentic iterations per request",
-  'Set {highlight}"tools": {"bash": false}{/highlight} to disable specific tools',
+  'Set {highlight}"tools": {"shell": false}{/highlight} to disable specific tools',
   'Set {highlight}"mcp_*": false{/highlight} to disable all tools from an MCP server',
   "Override global tool settings per agent configuration",
   'Set {highlight}"share": "auto"{/highlight} to automatically share all sessions',

packages/tui/src/routes/session/index.tsx

@@ -1622,7 +1622,7 @@ function ToolPart(props: { part: SessionMessageAssistantTool }) {
   return (
     <Show when={!shouldHide()}>
       <Switch>
-        <Match when={display() === "bash"}>
+        <Match when={display() === "shell"}>
           <Shell {...toolprops} />
         </Match>
         <Match when={display() === "glob"}>
@@ -1646,7 +1646,7 @@ function ToolPart(props: { part: SessionMessageAssistantTool }) {
         <Match when={display() === "edit"}>
           <Edit {...toolprops} />
         </Match>
-        <Match when={display() === "task"}>
+        <Match when={display() === "subagent"}>
           <Task {...toolprops} />
         </Match>
         <Match when={display() === "apply_patch"}>
@@ -2111,8 +2111,8 @@ function Task(props: ToolProps) {
       }}
     >
       {formatSubagentTitle(
-        Locale.titlecase(stringValue(props.input.subagent_type) ?? "General"),
-        description() ?? "Task",
+        Locale.titlecase(stringValue(props.input.agent) ?? stringValue(props.input.subagent_type) ?? "General"),
+        description() ?? "Subagent",
         props.metadata.background === true,
       )}
     </InlineTool>
@@ -2124,7 +2124,7 @@ export function formatSubagentToolcalls(count: number) {
 }
 
 export function formatSubagentTitle(agent: string, description: string, background: boolean) {
-  return `${agent} Task${background ? " (background)" : ""} — ${description}`
+  return `${agent} Subagent${background ? " (background)" : ""} — ${description}`
 }
 
 export function formatSubagentRetry(attempt: number, message: string) {
@@ -2402,7 +2402,7 @@ function numberValue(value: unknown) {
 }
 
 const toolDisplays = new Set([
-  "bash",
+  "shell",
   "glob",
   "read",
   "grep",
@@ -2410,7 +2410,7 @@ const toolDisplays = new Set([
   "websearch",
   "write",
   "edit",
-  "task",
+  "subagent",
   "apply_patch",
   "todowrite",
   "question",
@@ -2418,7 +2418,10 @@ const toolDisplays = new Set([
 ])
 
 export function toolDisplay(tool: string) {
-  return toolDisplays.has(tool) ? tool : "generic"
+  // Legacy transcripts recorded the shell tool as "bash" and the subagent tool as "task"; render
+  // them with the renamed views.
+  const normalized = tool === "bash" ? "shell" : tool === "task" ? "subagent" : tool
+  return toolDisplays.has(normalized) ? normalized : "generic"
 }
 
 function recordValue(value: unknown): Record<string, unknown> | undefined {

packages/tui/src/routes/session/permission.tsx

@@ -287,7 +287,7 @@ export function PermissionPrompt(props: { request: PermissionV2Request; director
               }
             }
 
-            if (permission === "bash") {
+            if (permission === "shell") {
               const command = typeof data.command === "string" ? data.command : ""
               return {
                 body: (
@@ -300,12 +300,17 @@ export function PermissionPrompt(props: { request: PermissionV2Request; director
               }
             }
 
-            if (permission === "task") {
-              const type = typeof data.subagent_type === "string" ? data.subagent_type : "Unknown"
+            if (permission === "subagent" || permission === "task") {
+              const agent =
+                typeof data.agent === "string"
+                  ? data.agent
+                  : typeof data.subagent_type === "string"
+                    ? data.subagent_type
+                    : "Unknown"
               const desc = typeof data.description === "string" ? data.description : ""
               return {
                 icon: "#",
-                title: `${Locale.titlecase(type)} Task`,
+                title: `${Locale.titlecase(agent)} Subagent`,
                 body: (
                   <Show when={desc}>
                     <box paddingLeft={1}>

packages/tui/test/cli/tui/__snapshots__/inline-tool-wrap-snapshot.test.tsx.snap

@@ -20,68 +20,3 @@ exports[`TUI inline tool wrapping snapshots expanded tool errors under the tool
    ✱ Grep "export const OPENCODE_DB|OPENCODE_DB|OPENCODE_DEV|Global\\.
      Path\\.data|data =" in packages/opencode/src (115 matches)"
 `;
-
-exports[`TUI inline tool wrapping keeps separation after a shell output block 1`] = `
-"
-
-  $ ls
-
-  file.ts
-
-   ✱ Grep "OPENCODE.*DB|database|sqlite|drizzle|dev.*db|data.
-     *dir|xdg|APPDATA" in packages/opencode/src (151 matches)
-   ✱ Glob "**/*db*" in packages/opencode (6 matches)
-   → Read packages/opencode/src/storage/db.ts [offset=1, limit=130]
-   → Read packages/opencode/src/index.ts [offset=1, limit=100]
-   ✱ Grep "export const OPENCODE_DB|OPENCODE_DB|OPENCODE_DEV|Global\\.
-     Path\\.data|data =" in packages/opencode/src (115 matches)"
-`;
-
-exports[`TUI inline tool wrapping keeps separation after a padded user message 1`] = `
-"
-  Check whether the next tool remains separated.
-
-
-   ✱ Grep "OPENCODE.*DB|database|sqlite|drizzle|dev.*db|data.
-     *dir|xdg|APPDATA" in packages/opencode/src (151 matches)
-   ✱ Glob "**/*db*" in packages/opencode (6 matches)
-   → Read packages/opencode/src/storage/db.ts [offset=1, limit=130]
-   → Read packages/opencode/src/index.ts [offset=1, limit=100]
-   ✱ Grep "export const OPENCODE_DB|OPENCODE_DB|OPENCODE_DEV|Global\\.
-     Path\\.data|data =" in packages/opencode/src (115 matches)"
-`;
-
-exports[`TUI inline tool wrapping separates after a multi-line task row 1`] = `
-"   ✱ Grep "Task" (2 matches)
-
-   ⠙ Explore Task — Inspect active task spacing
-
-   ✓ General Task — Confirm completed task spacing
-     ↳ 1 toolcall · 501ms
-
-   → Read src/cli/cmd/tui/routes/session/index.tsx"
-`;
-
-exports[`TUI inline tool wrapping separates a task row from a preceding inline detail 1`] = `
-"   → Read src/cli/cmd/tui/routes/session/index.tsx
-   ↳ Loaded src/cli/cmd/tui/routes/session/tools.tsx
-
-   ✓ Explore Task — Inspect active task spacing
-     ↳ 1 toolcall · 501ms"
-`;
-
-exports[`TUI inline tool wrapping separates an inline row from the previous assistant summary 1`] = `
-"   Build · Little Frank · 53.1s
-
-   ✓ Build Task — Review changes
-     ↳ 48 toolcalls · 1m 40s"
-`;
-
-exports[`TUI inline tool wrapping separates an inline row from the previous assistant error 1`] = `
-"│
-│  Managed inference requires an active Member plan
-│
-
-   ✓ Build Task — Review changes
-     ↳ 48 toolcalls · 1m 40s"
-`;

packages/tui/test/cli/tui/inline-tool-wrap-snapshot.test.tsx

@@ -104,7 +104,11 @@ async function renderFrame(component: () => JSX.Element, options: { width: numbe
 
 describe("TUI inline tool wrapping", () => {
   test("falls back for unknown tool names", () => {
-    expect(toolDisplay("bash")).toBe("bash")
+    expect(toolDisplay("shell")).toBe("shell")
+    expect(toolDisplay("subagent")).toBe("subagent")
+    // Legacy tool names normalize to their renamed views.
+    expect(toolDisplay("bash")).toBe("shell")
+    expect(toolDisplay("task")).toBe("subagent")
     expect(toolDisplay("plugin_tool")).toBe("generic")
   })
 
@@ -169,9 +173,9 @@ describe("TUI inline tool wrapping", () => {
   })
 
   test("keeps background state attached to the subagent identity", () => {
-    expect(formatSubagentTitle("Explore", "Inspect renderer", false)).toBe("Explore Task — Inspect renderer")
+    expect(formatSubagentTitle("Explore", "Inspect renderer", false)).toBe("Explore Subagent — Inspect renderer")
     expect(formatSubagentTitle("Explore", "Inspect renderer", true)).toBe(
-      "Explore Task (background) — Inspect renderer",
+      "Explore Subagent (background) — Inspect renderer",
     )
   })
 

specs/v2/plugin-session-cycle.prototype.ts

@@ -1,200 +0,0 @@
-/**
- * Prototype: plugin Session API dependency patterns.
- *
- * Run from repo root with:
- *
- *   bun specs/v2/plugin-session-cycle.prototype.ts
- *
- * The first case recreates the stripped-down cycle. The second case is the first
- * working version: PluginService stays in the location runtime, but it no longer
- * constructs PluginHost. A global PluginSupervisor constructs the host after it
- * can see both Session and a concrete location runtime.
- */
-
-type NodeName =
-  | "App"
-  | "Session"
-  | "LocationSession"
-  | "LocationServiceMap"
-  | "LocationRuntime"
-  | "InstanceState"
-  | "PluginService"
-  | "PluginHost"
-  | "PluginInternal"
-  | "PluginSupervisor"
-  | "SDK"
-  | "ToolDomain"
-
-type Graph = Readonly<Record<NodeName, readonly NodeName[]>>
-
-const empty: Graph = {
-  App: [],
-  Session: [],
-  LocationSession: [],
-  LocationServiceMap: [],
-  LocationRuntime: [],
-  InstanceState: [],
-  PluginService: [],
-  PluginHost: [],
-  PluginInternal: [],
-  PluginSupervisor: [],
-  SDK: [],
-  ToolDomain: [],
-}
-
-/**
- * RED: current shape if ctx.session is added to the host PluginService builds.
- *
- * Walkthrough:
- * - App needs Session and LocationServiceMap.
- * - Session needs LocationServiceMap for APIs like prompt/revert that route to a location.
- * - LocationServiceMap builds LocationRuntime.
- * - LocationRuntime builds PluginService.
- * - PluginService builds PluginHost.
- * - PluginHost now wants Session for ctx.session.
- */
-const currentPluginOwnsHost: Graph = {
-  ...empty,
-  App: ["Session", "LocationServiceMap"],
-  Session: ["LocationServiceMap"],
-  LocationServiceMap: ["LocationRuntime"],
-  LocationRuntime: ["PluginService", "PluginInternal", "ToolDomain"],
-  PluginService: ["PluginHost"],
-  PluginHost: ["Session", "ToolDomain"],
-  PluginInternal: ["PluginService"],
-}
-
-/**
- * GREEN v1: host construction moves out of PluginService.
- *
- * PluginService remains per-location, but is only lifecycle/scope ownership.
- * PluginSupervisor is app/global. It loads configured plugins for a location by:
- * - reading Session for ctx.session
- * - asking LocationServiceMap for that location runtime
- * - building PluginHost from those concrete capabilities
- * - asking the location PluginService to own the plugin scope
- *
- * LocationRuntime no longer needs PluginHost or PluginInternal while it is being
- * constructed, so Session can route through LocationServiceMap without looping
- * back into Session.
- */
-const supervisorOwnsHost: Graph = {
-  ...empty,
-  App: ["Session", "LocationServiceMap", "PluginSupervisor"],
-  Session: ["LocationServiceMap"],
-  LocationServiceMap: ["LocationRuntime"],
-  LocationRuntime: ["PluginService", "ToolDomain"],
-  PluginService: [],
-  PluginSupervisor: ["Session", "LocationServiceMap", "PluginHost", "PluginService"],
-  PluginHost: ["Session", "ToolDomain"],
-}
-
-/**
- * GREEN v2: split location-sensitive functions into a location service.
- *
- * Session is global data: create/get/list/messages/etc. It does not route into
- * LocationServiceMap. LocationSession is built inside the location runtime and
- * owns prompt/revert/other operations that touch location services.
- *
- * PluginHost can combine Session + LocationSession into one ctx.session surface,
- * while the graph still obeys: location services may depend on global services,
- * but global services do not depend on location services.
- */
-const locationSessionOperations: Graph = {
-  ...empty,
-  App: ["Session", "LocationServiceMap"],
-  LocationServiceMap: ["LocationRuntime"],
-  LocationRuntime: ["PluginService", "PluginHost", "LocationSession", "ToolDomain"],
-  PluginService: [],
-  PluginHost: ["Session", "LocationSession", "ToolDomain"],
-  LocationSession: ["Session", "ToolDomain"],
-}
-
-/**
- * GREEN but cursed: all services are global and read the active location from
- * InstanceState. This removes LocationServiceMap from the construction graph.
- *
- * The cost is that location correctness becomes ambient: every operation must
- * trust that InstanceState currently points at the session's location, or add
- * runtime assertions to catch a wrong ambient location.
- */
-const allGlobalInstanceState: Graph = {
-  ...empty,
-  App: ["InstanceState", "Session", "PluginService", "ToolDomain"],
-  Session: ["InstanceState"],
-  PluginService: ["PluginHost"],
-  PluginHost: ["Session", "ToolDomain", "InstanceState"],
-  ToolDomain: ["InstanceState"],
-}
-
-/**
- * GREEN v3: SDK as one plugin instance.
- *
- * The SDK does not install a location-specific plugin or write a global tool
- * registry. It receives the same PluginHost context as any plugin instance, and
- * its methods are wrappers over that host. When plugins are booted per location,
- * this SDK-backed plugin instance contributes normal location-local transforms.
- */
-const sdkAsPluginInstance: Graph = {
-  ...empty,
-  App: ["Session", "LocationServiceMap"],
-  LocationServiceMap: ["LocationRuntime"],
-  LocationRuntime: ["PluginService", "PluginHost", "LocationSession", "ToolDomain", "SDK"],
-  PluginService: [],
-  PluginHost: ["Session", "LocationSession", "ToolDomain"],
-  SDK: ["PluginHost"],
-  LocationSession: ["Session", "ToolDomain"],
-}
-
-expectCycle("current PluginService owns PluginHost with ctx.session", currentPluginOwnsHost)
-assertAcyclic("supervisor owns PluginHost; PluginService is lifecycle only", supervisorOwnsHost)
-assertAcyclic("location Session operations; globals do not route down", locationSessionOperations)
-assertAcyclic("all-global services with InstanceState ambient location", allGlobalInstanceState)
-assertAcyclic("SDK is a plugin instance that calls PluginHost", sdkAsPluginInstance)
-
-function expectCycle(name: string, graph: Graph) {
-  const cycle = findCycle(graph)
-  if (!cycle) throw new Error(`expected red but got green: ${name}`)
-  console.log(`red as expected: ${name}`)
-  console.log(`cycle: ${cycle.join(" -> ")}`)
-}
-
-function assertAcyclic(name: string, graph: Graph) {
-  const cycle = findCycle(graph)
-  if (!cycle) {
-    console.log(`green: ${name}`)
-    return
-  }
-  throw new Error(`red: ${name}\ncycle: ${cycle.join(" -> ")}`)
-}
-
-function findCycle(graph: Graph) {
-  const visiting = new Set<NodeName>()
-  const visited = new Set<NodeName>()
-  const stack: NodeName[] = []
-
-  const visit = (node: NodeName): NodeName[] | undefined => {
-    if (visiting.has(node)) return [...stack.slice(stack.indexOf(node)), node]
-    if (visited.has(node)) return
-
-    visiting.add(node)
-    stack.push(node)
-
-    for (const next of graph[node]) {
-      const cycle = visit(next)
-      if (cycle) return cycle
-    }
-
-    stack.pop()
-    visiting.delete(node)
-    visited.add(node)
-    return
-  }
-
-  for (const node of Object.keys(graph) as NodeName[]) {
-    const cycle = visit(node)
-    if (cycle) return cycle
-  }
-
-  return
-}

specs/v2/plugin-session-tools-plan.md

@@ -1,124 +0,0 @@
-# Plugin Session and Tool Architecture Prototype
-
-## Goal
-
-Let V2 plugins access a normal session API and define tools through plugin transforms without introducing application-vs-location tools or a `Session -> LocationServiceMap -> Plugin -> Session` construction cycle.
-
-## Recommended direction
-
-Keep location-scoped services. Split location-specific session behavior out of the global session data service.
-
-- `SessionV2.Service` remains global and owns session data APIs:
-  - `create`
-  - `get`
-  - `list`
-  - `messages`
-  - `message`
-  - `context`
-  - `events`
-  - `history`
-  - metadata-only updates such as `rename`, `switchAgent`, `switchModel`
-- Add a location-scoped session runtime service for behavior that touches location services:
-  - `prompt`
-  - `resume`
-  - `wait`
-  - `interrupt`
-  - `active`
-  - `revert.stage`
-  - `revert.clear`
-  - any future runner/filesystem/snapshot-coupled session operation
-
-This keeps the dependency rule simple:
-
-```txt
-global services do not call location services
-location services may call global services
-```
-
-## Why
-
-The current V2 shape becomes cyclic if `PluginHost` gets full `ctx.session` while `SessionV2.Service` depends on `LocationServiceMap`:
-
-```txt
-SessionV2
--> LocationServiceMap
--> LocationRuntime
--> PluginService
--> PluginHost
--> SessionV2
-```
-
-The prototype in `plugin-session-cycle.prototype.ts` recreates this red case and compares green alternatives.
-
-## Ideal plugin call site
-
-Plugin authors should not see the split. The host composes global session data plus location session runtime into one `ctx.session` API:
-
-```ts
-export const Plugin = define({
-  id: "subagent",
-  effect: Effect.fn(function* (ctx) {
-    yield* ctx.tool.transform((draft) => {
-      draft.set("subagent", tool({
-        description,
-        input: Input,
-        output: Output,
-        execute: Effect.fn(function* (input, call) {
-          const parent = yield* ctx.session.get(call.sessionID)
-          const child = yield* ctx.session.create({
-            parentID: parent.id,
-            title: input.description,
-            agent: input.agent,
-          })
-
-          yield* ctx.session.prompt({
-            sessionID: child.id,
-            prompt: { text: input.prompt },
-            resume: false,
-          })
-          yield* ctx.session.resume(child.id)
-          yield* ctx.session.wait(child.id)
-
-          return {
-            sessionID: child.id,
-            status: "completed",
-            output: yield* ctx.session.finalText(child.id),
-          }
-        }),
-      }))
-    })
-  }),
-})
-```
-
-## Tool model
-
-Tools stay location-scoped. There should not be public application tools or global tools.
-
-Plugins boot per location, and tools are contributed through transforms:
-
-```ts
-yield* ctx.tool.transform((draft) => {
-  draft.set("repo_summary", tool({ description, input, output, execute }))
-})
-```
-
-The SDK should be implemented as one plugin instance: it receives a plugin host/context internally, and SDK methods call the host methods. Therefore an SDK-registered tool is just a plugin tool transform applied as each location boots.
-
-## Concrete implementation slices
-
-1. Add `packages/core/src/session/runtime.ts` as a location node. **Implemented in this draft.**
-2. Move `prompt`, `resume`, `wait`, `interrupt`, `active`, and location-sensitive `revert` operations from `SessionV2.Service` into the runtime service. **Implemented in this draft for the new runtime path; old `SessionV2` entrypoints are left as compatibility stubs and should be removed once callers migrate.**
-3. Update server route handlers to route location-sensitive requests at the API boundary by resolving the session location and providing that location runtime. **Implemented in this draft.**
-4. Add `ctx.session` to `PluginHost` by composing `SessionV2.Service` and the location session runtime. **Implemented in this draft.**
-5. Add public plugin `ctx.tool.transform` types and adapt it to the existing canonical core `Tool.make` representation.
-6. Convert `ToolRegistry` registration to transform/rebuild semantics.
-7. Port `subagent` to a built-in plugin that registers a normal location tool.
-8. Remove `ApplicationTools` once no built-in tool requires process-global registration.
-
-## Invariants to preserve
-
-- A tool materialization snapshots executable tool identity; stale calls fail.
-- Tool output bounding remains centralized in `ToolRegistry.Materialization.settle`.
-- Location session runtime asserts that the target session belongs to the current location before running location-sensitive operations.
-- Public HTTP/SDK API shape does not need to change.

specs/v2/session.md

@@ -135,7 +135,7 @@ Status: `complete` is usable in the native V2 path, `partial` covers only part o
 | Durable Context Source     | Selected-agent available skill guidance and skill-body loading           | partial  | Guidance and body exposure are permission-filtered; remove globally denied skill definitions during request-time tool materialization. |
 | Per-turn request assembly  | Placement, selected model, chronological history, and canonical lowering | complete | None.                                                                                                                                  |
 | Per-turn request assembly  | Selected agent, agent prompt, and effective permissions                  | partial  | V2 uses selected-agent permissions for skill guidance and tool authorization; still apply the agent system prompt and request policy.  |
-| Per-turn request assembly  | Provider/model-specific base instructions                                | missing  | Select the provider-family baseline unless the effective agent overrides it.                                                           |
+| Per-turn request assembly  | Provider/model-specific base instructions                                | complete | Native V2 selects the provider-family baseline unless the effective agent overrides it.                                                |
 | Per-turn request assembly  | Policy-filtered built-in, MCP, plugin, and structured-output tools       | partial  | Materialize definitions for the effective agent and request.                                                                           |
 | Per-turn request assembly  | Per-prompt system text and tool overrides                                | missing  | Design admission and durable replay semantics before exposing them.                                                                    |
 | Per-turn request assembly  | Steering, plan/build-switch, and final-step reminders                    | missing  | Add only reminders whose behavior remains part of V2.                                                                                  |