furyhawk/pydantic-ai-harness
1
0
Fork 0
mirror of https://github.com/pydantic/pydantic-ai-harness.git synced 2026-07-21 10:55:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove Temporal section from Code Mode README

Browse Source 
Drop the Temporal durable-execution docs while the pydantic_monty passthrough
question is unresolved, to avoid documenting a setup that is in flux.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Aditya Vardhan
2026-07-13 21:05:03 +05:30
co-authored by Claude Opus 4.8
parent 73ca5f49df
commit bf9d8eff12
1 changed files with 0 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files
pydantic_ai_harness/code_mode/README.md
-40
View File
@@ -154,46 +154,6 @@ The last expression in the code snippet is automatically captured as the return
State persists between `run_code` calls within the same agent run -- variables, imports, and function definitions carry over. Pass `restart: true` in the tool call to reset state.
## Temporal durable execution
Code mode works with Pydantic AI's Temporal integration, but the Temporal worker must pass `pydantic_monty`
through its workflow import sandbox. Monty's native module manages the subprocess pool used to execute code. Passing
the module through keeps that pool outside the per-workflow import sandbox without disabling sandboxing for the rest
of the workflow.
Add a custom workflow runner to the `Worker` that runs the durable agent:
```python
from pydantic_ai.durable_exec.temporal import PydanticAIPlugin
from temporalio.client import Client
from temporalio.worker import Worker
from temporalio.worker.workflow_sandbox import SandboxedWorkflowRunner, SandboxRestrictions
workflow_runner = SandboxedWorkflowRunner(
restrictions=SandboxRestrictions.default.with_passthrough_modules('pydantic_monty')
)
client = await Client.connect(
'localhost:7233',
plugins=[PydanticAIPlugin()],
)
async with Worker(
client,
task_queue='code-mode',
workflows=[CodeModeWorkflow],
workflow_runner=workflow_runner,
):
...
```
`PydanticAIPlugin` merges its normal passthrough modules into the runner's restrictions. The workflow remains
sandboxed, and the same runner can be passed to Temporal's `Replayer`.
Monty code runs again when Temporal replays the workflow. Keep external state access out of the code sandbox: do not
connect `os_access` or `mount` to changing host state. Put filesystem, network, clock, and environment access in tools
so `TemporalAgent` can execute those calls as activities and record their results in workflow history.
## Observability
Nested tool calls inside `run_code` produce their own spans when instrumented with [Logfire](https://pydantic.dev/logfire) or any OpenTelemetry backend. The `run_code` tool return includes metadata with all nested calls: