furyhawk/pydantic-ai-harness
1
0
Fork 0
mirror of https://github.com/pydantic/pydantic-ai-harness.git synced 2026-07-21 02:45:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

Rename package to pydantic-ai-harness (#207)

Browse Source 
- Package name: pydantic-harness -> pydantic-ai-harness
- Python module: pydantic_harness -> pydantic_ai_harness
- Display name: Pydantic Harness -> Pydantic AI Harness
- All URLs updated to github.com/pydantic/pydantic-ai-harness

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Douwe Maan
2026-04-13 14:33:28 -06:00
committed by GitHub
co-authored by Claude Opus 4.6
parent 8cb9941468
commit 987b7312d5
17 changed files with 140 additions and 140 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/ISSUE_TEMPLATE/bug-report.yml
+1 -1
View File
@@ -28,7 +28,7 @@ body:
- type: input
id: version
attributes:
label: pydantic-harness version
label: pydantic-ai-harness version
placeholder: e.g., 0.1.0
validations:
required: true
.github/ISSUE_TEMPLATE/capability-request.yml
+1 -1
View File
@@ -1,5 +1,5 @@
name: Capability Request
description: Propose a new capability for pydantic-harness
description: Propose a new capability for pydantic-ai-harness
labels: ['capability', 'needs:triage']
body:
- type: input
AGENTS.md
+3 -3
View File
@@ -1,4 +1,4 @@
# Pydantic Harness
# Pydantic AI Harness
## Local setup
@@ -16,7 +16,7 @@ Fill in the absolute path where the `capabilities` branch of `pydantic-ai` is ch
## Vocabulary
- **Capability**: an `AbstractCapability` subclass that bundles tools, hooks, instructions, and model settings into a reusable unit. This is the core abstraction of pydantic-harness
- **Capability**: an `AbstractCapability` subclass that bundles tools, hooks, instructions, and model settings into a reusable unit. This is the core abstraction of pydantic-ai-harness
- **Hook**: a lifecycle method on `AbstractCapability` that intercepts agent graph execution (e.g. `before_model_request`, `wrap_run`, `after_tool_execute`)
- **Toolset**: a collection of tools that a capability can provide to the agent
- **Guard**: a type of capability that validates inputs/outputs or controls tool access (e.g. `InputGuardrail`, `CostGuard`)
@@ -71,7 +71,7 @@ Always run `make lint && make typecheck && make test` before committing.
## File structure
```
pydantic_harness/
pydantic_ai_harness/
__init__.py # public API re-exports
<capability>/ # each capability gets its own package
__init__.py # public exports for the capability
README.md
+48 -48
View File
@@ -1,9 +1,9 @@
# Pydantic Harness
# Pydantic AI Harness
[![CI](https://github.com/pydantic/pydantic-harness/actions/workflows/main.yml/badge.svg?event=push)](https://github.com/pydantic/pydantic-harness/actions/workflows/main.yml?query=branch%3Amain)
[![PyPI](https://img.shields.io/pypi/v/pydantic-harness.svg)](https://pypi.python.org/pypi/pydantic-harness)
[![versions](https://img.shields.io/pypi/pyversions/pydantic-harness.svg)](https://github.com/pydantic/pydantic-harness)
[![license](https://img.shields.io/github/license/pydantic/pydantic-harness.svg)](https://github.com/pydantic/pydantic-harness/blob/main/LICENSE)
[![CI](https://github.com/pydantic/pydantic-ai-harness/actions/workflows/main.yml/badge.svg?event=push)](https://github.com/pydantic/pydantic-ai-harness/actions/workflows/main.yml?query=branch%3Amain)
[![PyPI](https://img.shields.io/pypi/v/pydantic-ai-harness.svg)](https://pypi.python.org/pypi/pydantic-ai-harness)
[![versions](https://img.shields.io/pypi/pyversions/pydantic-ai-harness.svg)](https://github.com/pydantic/pydantic-ai-harness)
[![license](https://img.shields.io/github/license/pydantic/pydantic-ai-harness.svg)](https://github.com/pydantic/pydantic-ai-harness/blob/main/LICENSE)
**The batteries for your [Pydantic AI](https://ai.pydantic.dev/) agent.**
@@ -11,7 +11,7 @@
Pydantic AI's [capabilities](https://ai.pydantic.dev/capabilities/) and [hooks](https://ai.pydantic.dev/hooks/) API is how you give an agent its harness -- bundles of tools, lifecycle hooks, instructions, and model settings that extend what the agent can do without any framework changes.
**Pydantic Harness** is the official capability library for Pydantic AI, maintained by the [Pydantic AI](https://github.com/pydantic/pydantic-ai) team. Pydantic AI core ships capabilities that require model or framework support, and capabilities fundamental to every agent -- [web search](https://ai.pydantic.dev/capabilities/#provider-adaptive-tools), [tool search](https://ai.pydantic.dev/deferred-tools/), [thinking](https://ai.pydantic.dev/capabilities/#thinking). Everything else lives here: standalone building blocks you pick and choose to turn your agent into a coding agent, a research assistant, or anything else. This is also where new capabilities start -- as they stabilize and prove themselves broadly essential, they can graduate into core.
**Pydantic AI Harness** is the official capability library for Pydantic AI, maintained by the [Pydantic AI](https://github.com/pydantic/pydantic-ai) team. Pydantic AI core ships capabilities that require model or framework support, and capabilities fundamental to every agent -- [web search](https://ai.pydantic.dev/capabilities/#provider-adaptive-tools), [tool search](https://ai.pydantic.dev/deferred-tools/), [thinking](https://ai.pydantic.dev/capabilities/#thinking). Everything else lives here: standalone building blocks you pick and choose to turn your agent into a coding agent, a research assistant, or anything else. This is also where new capabilities start -- as they stabilize and prove themselves broadly essential, they can graduate into core.
The [capability matrix](#capability-matrix) tracks where we are. [Tell us what to prioritize.](#help-us-prioritize)
@@ -20,13 +20,13 @@ The [capability matrix](#capability-matrix) tracks where we are. [Tell us what t
## Installation
```bash
uv add pydantic-harness
uv add pydantic-ai-harness
```
Extras for specific capabilities:
```bash
uv add "pydantic-harness[code-mode]" # CodeMode (adds the Monty sandbox)
uv add "pydantic-ai-harness[code-mode]" # CodeMode (adds the Monty sandbox)
```
Requires Python 3.10+ and `pydantic-ai-slim>=1.80.0`.
@@ -37,7 +37,7 @@ Requires Python 3.10+ and `pydantic-ai-slim>=1.80.0`.
import logfire
from pydantic_ai import Agent
from pydantic_ai.capabilities import MCP # from the core pydantic-ai package
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
logfire.configure()
logfire.instrument_pydantic_ai()
@@ -50,56 +50,56 @@ agent = Agent(
],
)
result = agent.run_sync('Rank the open PRs on pydantic/pydantic-harness by thumbs-up reactions. Which 5 should we merge first?')
result = agent.run_sync('Rank the open PRs on pydantic/pydantic-ai-harness by thumbs-up reactions. Which 5 should we merge first?')
print(result.output)
```
[`MCP`](https://ai.pydantic.dev/capabilities/#provider-adaptive-tools) (from the core `pydantic-ai` package) connects your agent to any MCP server -- here, [GitHub's official MCP server](https://github.com/github/github-mcp-server).
[`CodeMode`](pydantic_harness/code_mode/) wraps all tools into a single `run_code` tool powered by our [Monty](https://github.com/pydantic/monty) sandbox, so the model can orchestrate multiple tool calls with Python code instead of one model round-trip per call.
[`CodeMode`](pydantic_ai_harness/code_mode/) wraps all tools into a single `run_code` tool powered by our [Monty](https://github.com/pydantic/monty) sandbox, so the model can orchestrate multiple tool calls with Python code instead of one model round-trip per call.
[`logfire`](https://pydantic.dev/logfire) gives you a trace for every agent run. With CodeMode, you can see the `run_code` span with each nested tool call as a child span -- making it easy to debug what the model's code actually did. See the [Pydantic AI Logfire docs](https://ai.pydantic.dev/logfire/) for setup details.
## Capability matrix
We studied leading coding agents, agent frameworks, and Claw-style assistants to map every capability area that matters for production agents. Each one is tracked as an [issue](https://github.com/pydantic/pydantic-harness/issues) in this repo.
We studied leading coding agents, agent frameworks, and Claw-style assistants to map every capability area that matters for production agents. Each one is tracked as an [issue](https://github.com/pydantic/pydantic-ai-harness/issues) in this repo.
**Vote on whatever is linked in the Status column** -- PRs if we're actively building it, issues if it's planned -- to help us decide what to work on next.
| Category | Capability | Description | Status | Community&nbsp;alternatives |
|---|---|---|---|---|
| **Tools &&nbsp;execution** | **Code mode** | Sandboxed Python execution via [Monty](https://github.com/pydantic/monty) -- one `run_code` call replaces N tool calls | :white_check_mark: [Docs](pydantic_harness/code_mode/) | |
| **Tools &&nbsp;execution** | **Code mode** | Sandboxed Python execution via [Monty](https://github.com/pydantic/monty) -- one `run_code` call replaces N tool calls | :white_check_mark: [Docs](pydantic_ai_harness/code_mode/) | |
| | **Tool search** | Progressive tool discovery for large tool sets | :white_check_mark: [Pydantic&nbsp;AI](https://ai.pydantic.dev/deferred-tools/) | |
| | **File system** | Read, write, edit, search files with path traversal prevention | :construction: [PR&nbsp;#177](https://github.com/pydantic/pydantic-harness/pull/177) | [pydantic-ai-backend](https://github.com/vstorm-co/pydantic-ai-backend) (vstorm&#8209;co) |
| | **Shell** | Execute commands with allowlists, denylists, and timeouts | :construction: [PR&nbsp;#177](https://github.com/pydantic/pydantic-harness/pull/177) | [pydantic-ai-backend](https://github.com/vstorm-co/pydantic-ai-backend) (vstorm&#8209;co) |
| | **Repo context injection** | Auto-load CLAUDE.md/AGENTS.md and repo structure | :construction: [PR&nbsp;#175](https://github.com/pydantic/pydantic-harness/pull/175) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Verification loop** | Run tests after edits, auto-fix failures | :construction: [PR&nbsp;#169](https://github.com/pydantic/pydantic-harness/pull/169) | |
| **Context management** | **Sliding window** | Trim conversation history to stay within token limits | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Context compaction** | LLM-powered summarization of older messages | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Limit warnings** | Warn agent before hitting context/iteration limits | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Tool output management** | Truncate, summarize, or spill large tool outputs | :construction: [PR&nbsp;#185](https://github.com/pydantic/pydantic-harness/pull/185) | |
| | **System reminders** | Inject periodic reminders to counteract instruction drift | :construction: [PR&nbsp;#181](https://github.com/pydantic/pydantic-harness/pull/181) | |
| **Memory &&nbsp;persistence** | **Memory** | Persistent key-value memory across sessions | :construction: [PR&nbsp;#179](https://github.com/pydantic/pydantic-harness/pull/179) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Session persistence** | Save and restore full conversation state | :construction: [PR&nbsp;#176](https://github.com/pydantic/pydantic-harness/pull/176) | |
| | **Checkpointing** | Save, rewind, and fork conversation state | :memo: [#196](https://github.com/pydantic/pydantic-harness/issues/196) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| **Agent orchestration** | **Sub-agents** | Delegate subtasks to specialized child agents | :construction: [PR&nbsp;#178](https://github.com/pydantic/pydantic-harness/pull/178) | [subagents-pydantic-ai](https://github.com/vstorm-co/subagents-pydantic-ai) (vstorm&#8209;co) |
| | **Skills** | Progressive tool loading -- search, activate, deactivate | :construction: [PR&nbsp;#183](https://github.com/pydantic/pydantic-harness/pull/183) | [pydantic-ai-skills](https://github.com/DougTrajano/pydantic-ai-skills) (DougTrajano), [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Planning** | Break complex tasks into structured plans before execution | :construction: [PR&nbsp;#180](https://github.com/pydantic/pydantic-harness/pull/180) | |
| | **Task tracking** | Track tasks, subtasks, and dependencies | :memo: [#65](https://github.com/pydantic/pydantic-harness/issues/65) | [pydantic-ai-todo](https://github.com/vstorm-co/pydantic-ai-todo) (vstorm&#8209;co) |
| | **Teams** | Multi-agent teams with shared state and message bus | :memo: [#195](https://github.com/pydantic/pydantic-harness/issues/195) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| **Safety &&nbsp;guardrails** | **Input guardrails** | Validate user input before the agent run starts | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Output guardrails** | Validate model output after the run completes | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Cost/token budgets** | Enforce token and cost limits per run | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Tool access control** | Block tools or require approval before execution | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Async guardrails** | Run validation concurrently with model requests | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Secret masking** | Detect and redact secrets in agent I/O | :construction: [PR&nbsp;#172](https://github.com/pydantic/pydantic-harness/pull/172) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Approval workflows** | Require human approval for sensitive operations | :construction: [PR&nbsp;#173](https://github.com/pydantic/pydantic-harness/pull/173) | [Pydantic&nbsp;AI](https://ai.pydantic.dev/deferred-tools/#human-in-the-loop-tool-approval) (built&#8209;in) |
| | **Tool budget** | Limit total tool calls or cost per run | :construction: [PR&nbsp;#168](https://github.com/pydantic/pydantic-harness/pull/168) | |
| **Reliability** | **Stuck loop detection** | Detect and break out of repetitive agent loops | :construction: [PR&nbsp;#186](https://github.com/pydantic/pydantic-harness/pull/186) | |
| | **Tool error recovery** | Retry failed tool calls with backoff and budget | :construction: [PR&nbsp;#171](https://github.com/pydantic/pydantic-harness/pull/171) | |
| | **Tool orphan repair** | Fix orphaned tool calls in conversation history | :construction: [PR&nbsp;#184](https://github.com/pydantic/pydantic-harness/pull/184) | |
| **Reasoning** | **Adaptive reasoning** | Adjust thinking effort based on task complexity | :construction: [PR&nbsp;#174](https://github.com/pydantic/pydantic-harness/pull/174) | |
| | **Current time** | Inject current date/time into system prompt | :construction: [PR&nbsp;#170](https://github.com/pydantic/pydantic-harness/pull/170) | |
| | **File system** | Read, write, edit, search files with path traversal prevention | :construction: [PR&nbsp;#177](https://github.com/pydantic/pydantic-ai-harness/pull/177) | [pydantic-ai-backend](https://github.com/vstorm-co/pydantic-ai-backend) (vstorm&#8209;co) |
| | **Shell** | Execute commands with allowlists, denylists, and timeouts | :construction: [PR&nbsp;#177](https://github.com/pydantic/pydantic-ai-harness/pull/177) | [pydantic-ai-backend](https://github.com/vstorm-co/pydantic-ai-backend) (vstorm&#8209;co) |
| | **Repo context injection** | Auto-load CLAUDE.md/AGENTS.md and repo structure | :construction: [PR&nbsp;#175](https://github.com/pydantic/pydantic-ai-harness/pull/175) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Verification loop** | Run tests after edits, auto-fix failures | :construction: [PR&nbsp;#169](https://github.com/pydantic/pydantic-ai-harness/pull/169) | |
| **Context management** | **Sliding window** | Trim conversation history to stay within token limits | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-ai-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Context compaction** | LLM-powered summarization of older messages | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-ai-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Limit warnings** | Warn agent before hitting context/iteration limits | :construction: [PR&nbsp;#191](https://github.com/pydantic/pydantic-ai-harness/pull/191) | [summarization-pydantic-ai](https://github.com/vstorm-co/summarization-pydantic-ai) (vstorm&#8209;co) |
| | **Tool output management** | Truncate, summarize, or spill large tool outputs | :construction: [PR&nbsp;#185](https://github.com/pydantic/pydantic-ai-harness/pull/185) | |
| | **System reminders** | Inject periodic reminders to counteract instruction drift | :construction: [PR&nbsp;#181](https://github.com/pydantic/pydantic-ai-harness/pull/181) | |
| **Memory &&nbsp;persistence** | **Memory** | Persistent key-value memory across sessions | :construction: [PR&nbsp;#179](https://github.com/pydantic/pydantic-ai-harness/pull/179) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Session persistence** | Save and restore full conversation state | :construction: [PR&nbsp;#176](https://github.com/pydantic/pydantic-ai-harness/pull/176) | |
| | **Checkpointing** | Save, rewind, and fork conversation state | :memo: [#196](https://github.com/pydantic/pydantic-ai-harness/issues/196) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| **Agent orchestration** | **Sub-agents** | Delegate subtasks to specialized child agents | :construction: [PR&nbsp;#178](https://github.com/pydantic/pydantic-ai-harness/pull/178) | [subagents-pydantic-ai](https://github.com/vstorm-co/subagents-pydantic-ai) (vstorm&#8209;co) |
| | **Skills** | Progressive tool loading -- search, activate, deactivate | :construction: [PR&nbsp;#183](https://github.com/pydantic/pydantic-ai-harness/pull/183) | [pydantic-ai-skills](https://github.com/DougTrajano/pydantic-ai-skills) (DougTrajano), [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| | **Planning** | Break complex tasks into structured plans before execution | :construction: [PR&nbsp;#180](https://github.com/pydantic/pydantic-ai-harness/pull/180) | |
| | **Task tracking** | Track tasks, subtasks, and dependencies | :memo: [#65](https://github.com/pydantic/pydantic-ai-harness/issues/65) | [pydantic-ai-todo](https://github.com/vstorm-co/pydantic-ai-todo) (vstorm&#8209;co) |
| | **Teams** | Multi-agent teams with shared state and message bus | :memo: [#195](https://github.com/pydantic/pydantic-ai-harness/issues/195) | [pydantic-deep](https://github.com/vstorm-co/pydantic-deepagents) (vstorm&#8209;co) |
| **Safety &&nbsp;guardrails** | **Input guardrails** | Validate user input before the agent run starts | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-ai-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Output guardrails** | Validate model output after the run completes | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-ai-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Cost/token budgets** | Enforce token and cost limits per run | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-ai-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Tool access control** | Block tools or require approval before execution | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-ai-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Async guardrails** | Run validation concurrently with model requests | :construction: [PR&nbsp;#182](https://github.com/pydantic/pydantic-ai-harness/pull/182) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Secret masking** | Detect and redact secrets in agent I/O | :construction: [PR&nbsp;#172](https://github.com/pydantic/pydantic-ai-harness/pull/172) | [pydantic-ai-shields](https://github.com/vstorm-co/pydantic-ai-shields) (vstorm&#8209;co) |
| | **Approval workflows** | Require human approval for sensitive operations | :construction: [PR&nbsp;#173](https://github.com/pydantic/pydantic-ai-harness/pull/173) | [Pydantic&nbsp;AI](https://ai.pydantic.dev/deferred-tools/#human-in-the-loop-tool-approval) (built&#8209;in) |
| | **Tool budget** | Limit total tool calls or cost per run | :construction: [PR&nbsp;#168](https://github.com/pydantic/pydantic-ai-harness/pull/168) | |
| **Reliability** | **Stuck loop detection** | Detect and break out of repetitive agent loops | :construction: [PR&nbsp;#186](https://github.com/pydantic/pydantic-ai-harness/pull/186) | |
| | **Tool error recovery** | Retry failed tool calls with backoff and budget | :construction: [PR&nbsp;#171](https://github.com/pydantic/pydantic-ai-harness/pull/171) | |
| | **Tool orphan repair** | Fix orphaned tool calls in conversation history | :construction: [PR&nbsp;#184](https://github.com/pydantic/pydantic-ai-harness/pull/184) | |
| **Reasoning** | **Adaptive reasoning** | Adjust thinking effort based on task complexity | :construction: [PR&nbsp;#174](https://github.com/pydantic/pydantic-ai-harness/pull/174) | |
| | **Current time** | Inject current date/time into system prompt | :construction: [PR&nbsp;#170](https://github.com/pydantic/pydantic-ai-harness/pull/170) | |
> Packages by [vstorm-co](https://github.com/vstorm-co) are endorsed by the Pydantic AI team. We're working with them to upstream some of their implementations into this repo.
@@ -107,7 +107,7 @@ We studied leading coding agents, agent frameworks, and Claw-style assistants to
**Vote on whatever is linked in the Status column above.** If there's a PR, vote on the PR -- it means we're actively building it. If there's only an issue, vote on the issue.
Want something that's not on the list? [Open a capability request](https://github.com/pydantic/pydantic-harness/issues/new?template=capability-request.yml).
Want something that's not on the list? [Open a capability request](https://github.com/pydantic/pydantic-ai-harness/issues/new?template=capability-request.yml).
## Build your own
@@ -119,11 +119,11 @@ Want something that's not on the list? [Open a capability request](https://githu
We welcome capability contributions. Here's how:
1. **Start with an issue.** [Open a capability request](https://github.com/pydantic/pydantic-harness/issues/new?template=capability-request.yml) describing the behavior you want. This lets us discuss the approach and priority before code is written -- we can close an approach without closing the problem.
1. **Start with an issue.** [Open a capability request](https://github.com/pydantic/pydantic-ai-harness/issues/new?template=capability-request.yml) describing the behavior you want. This lets us discuss the approach and priority before code is written -- we can close an approach without closing the problem.
2. **Then open a PR.** Once the issue exists, you're welcome to open a PR with an implementation. Link the issue in your PR. We review based on community interest -- upvotes on both the issue and PR count.
3. **Don't chase green CI.** Get the approach working, then let us know. We'll take it from there -- we may push to your branch, rewrite, or open a follow-up PR. You'll be credited as the original author. (See the [Pydantic AI contributing guide](https://github.com/pydantic/pydantic-ai/blob/main/CONTRIBUTING.md).)
> **Note**: PRs that modify `pyproject.toml` or `uv.lock` from non-team members are auto-closed by CI to prevent supply chain risk. If you need a new dependency, [open an issue](https://github.com/pydantic/pydantic-harness/issues/new).
> **Note**: PRs that modify `pyproject.toml` or `uv.lock` from non-team members are auto-closed by CI to prevent supply chain risk. If you need a new dependency, [open an issue](https://github.com/pydantic/pydantic-ai-harness/issues/new).
### Development
@@ -138,14 +138,14 @@ make testcov # pytest with 100% branch coverage
## Version policy
Pydantic Harness uses **0.x versioning** to signal that APIs are still stabilizing. During 0.x:
Pydantic AI Harness uses **0.x versioning** to signal that APIs are still stabilizing. During 0.x:
- **Minor releases** (0.1 → 0.2) may include breaking changes — renamed parameters, changed defaults, restructured APIs. As the library grows, especially as capabilities gain provider-native support (starting as a local implementation, then auto-switching to the provider's built-in API when available), we may need to reshape APIs we couldn't fully anticipate in the initial design.
- **Patch releases** (0.1.0 → 0.1.1) will not intentionally break existing behavior.
- **All breaking changes** are documented in release notes with migration guidance.
- Where practical, we'll keep the previous behavior available under a deprecated name or configuration option before removing it.
This is why Pydantic Harness is a separate package from [Pydantic AI](https://github.com/pydantic/pydantic-ai), which has a [stricter version policy](https://ai.pydantic.dev/version-policy/). As the core capabilities stabilize, we'll move toward 1.0 with stability guarantees to match.
This is why Pydantic AI Harness is a separate package from [Pydantic AI](https://github.com/pydantic/pydantic-ai), which has a [stricter version policy](https://ai.pydantic.dev/version-policy/). As the core capabilities stabilize, we'll move toward 1.0 with stability guarantees to match.
## Pydantic AI references
pydantic_harness/__init__.py → pydantic_ai_harness/__init__.py
+1 -1
View File
@@ -1,4 +1,4 @@
"""Agent harness for composable, reusable AI agent capabilities, for Pydantic AI."""
"""The batteries for your Pydantic AI agent -- the official capability library."""
from typing import TYPE_CHECKING
pydantic_harness/code_mode/README.md → pydantic_ai_harness/code_mode/README.md
+4 -4
View File
@@ -21,7 +21,7 @@ Standard tool calling requires one model round-trip per tool call. An agent that
```python
from pydantic_ai import Agent
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
agent = Agent('anthropic:claude-sonnet-4-6', capabilities=[CodeMode()])
@@ -56,7 +56,7 @@ tokyo_c = await convert_temp(fahrenheit=tokyo['temp_f'])
Code mode requires the Monty sandbox:
```bash
uv add "pydantic-harness[code-mode]"
uv add "pydantic-ai-harness[code-mode]"
```
## Selective tool sandboxing
@@ -81,7 +81,7 @@ Tools that match the selector are wrapped inside `run_code`. Non-matching tools
```python
from pydantic_ai import Agent
from pydantic_ai.toolsets import FunctionToolset
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
search_tools = FunctionToolset(tools=[search, fetch]).with_metadata(code_mode=True)
@@ -149,7 +149,7 @@ capabilities:
```python
from pydantic_ai import Agent
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
agent = Agent.from_file('agent.yaml', custom_capability_types=[CodeMode])
result = agent.run_sync('...')
pydantic_ai_harness/code_mode/__init__.py
+6
View File
@@ -0,0 +1,6 @@
"""Code mode capability: route tool calls through a sandboxed Python environment."""
from pydantic_ai_harness.code_mode._capability import CodeMode
from pydantic_ai_harness.code_mode._toolset import CodeModeToolset
__all__ = ['CodeMode', 'CodeModeToolset']
pydantic_harness/code_mode/_capability.py → pydantic_ai_harness/code_mode/_capability.py
+2 -2
View File
@@ -9,7 +9,7 @@ from pydantic_ai.capabilities import AbstractCapability, CapabilityOrdering
from pydantic_ai.capabilities._tool_search import ToolSearch as _ToolSearch
from pydantic_ai.tools import AgentDepsT, ToolSelector
from pydantic_harness.code_mode._toolset import CodeModeToolset
from pydantic_ai_harness.code_mode._toolset import CodeModeToolset
@dataclass
@@ -26,7 +26,7 @@ class CodeMode(AbstractCapability[AgentDepsT]):
```python
from pydantic_ai import Agent
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
# Sandbox all tools
agent = Agent('openai:gpt-5', capabilities=[CodeMode()])
pydantic_harness/code_mode/_toolset.py → pydantic_ai_harness/code_mode/_toolset.py
+1 -1
View File
@@ -41,7 +41,7 @@ try:
)
except ImportError as _import_error: # pragma: no cover
raise ImportError(
'pydantic-monty is required for CodeMode. Install it with: pip install "pydantic-harness[code-mode]"'
'pydantic-monty is required for CodeMode. Install it with: pip install "pydantic-ai-harness[code-mode]"'
) from _import_error
from typing_extensions import NotRequired, TypedDict
pydantic_harness/py.typed → pydantic_ai_harness/py.typed
View File
pydantic_harness/code_mode/__init__.py
-6
View File
@@ -1,6 +0,0 @@
"""Code mode capability: route tool calls through a sandboxed Python environment."""
from pydantic_harness.code_mode._capability import CodeMode
from pydantic_harness.code_mode._toolset import CodeModeToolset
__all__ = ['CodeMode', 'CodeModeToolset']
pyproject.toml
+8 -8
View File
@@ -3,7 +3,7 @@ requires = ['hatchling', 'uv-dynamic-versioning>=0.7.0']
build-backend = 'hatchling.build'
[project]
name = 'pydantic-harness'
name = 'pydantic-ai-harness'
dynamic = ['version']
description = 'The batteries for your Pydantic AI agent'
readme = 'README.md'
@@ -43,13 +43,13 @@ dbos = [
]
[project.urls]
Homepage = 'https://github.com/pydantic/pydantic-harness'
Source = 'https://github.com/pydantic/pydantic-harness'
Issues = 'https://github.com/pydantic/pydantic-harness/issues'
Homepage = 'https://github.com/pydantic/pydantic-ai-harness'
Source = 'https://github.com/pydantic/pydantic-ai-harness'
Issues = 'https://github.com/pydantic/pydantic-ai-harness/issues'
[dependency-groups]
dev = [
'pydantic-harness[code-mode]',
'pydantic-ai-harness[code-mode]',
'pytest',
'anyio[trio]',
'pytest-anyio',
@@ -74,7 +74,7 @@ bump = true
[tool.hatch.build.targets.wheel]
packages = ['pydantic_harness']
packages = ['pydantic_ai_harness']
[tool.ruff]
line-length = 120
@@ -119,10 +119,10 @@ anyio_mode = 'auto'
[tool.coverage.run]
branch = true
source = ['pydantic_harness', 'tests']
source = ['pydantic_ai_harness', 'tests']
[tool.coverage.paths]
source = ['.', '/home/runner/work/pydantic-harness/pydantic-harness']
source = ['.', '/home/runner/work/pydantic-ai-harness/pydantic-ai-harness']
[tool.coverage.report]
fail_under = 100
tests/_code_mode/test_code_mode.py
+3 -3
View File
@@ -26,9 +26,9 @@ from pydantic_ai.usage import RunUsage
from pydantic_core import SchemaValidator, core_schema
from typing_extensions import TypedDict
from pydantic_harness import CodeMode
from pydantic_harness.code_mode import CodeModeToolset
from pydantic_harness.code_mode._toolset import ( # pyright: ignore[reportPrivateUsage]
from pydantic_ai_harness import CodeMode
from pydantic_ai_harness.code_mode import CodeModeToolset
from pydantic_ai_harness.code_mode._toolset import ( # pyright: ignore[reportPrivateUsage]
_SEARCH_TOOLS_MODIFIER,
_TOOL_SEARCH_ADDENDUM,
_PrintCapture,
tests/_code_mode/test_dbos.py
+2 -2
View File
@@ -25,7 +25,7 @@ from pydantic_ai.messages import ModelRequest, ModelResponse, TextPart, ToolCall
from pydantic_ai.models.function import AgentInfo, FunctionModel
from pydantic_ai.toolsets.function import FunctionToolset
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
# ---------------------------------------------------------------------------
# Fixtures
@@ -36,7 +36,7 @@ from pydantic_harness import CodeMode
def dbos_instance(tmp_path_factory: pytest.TempPathFactory) -> Generator[DBOS, Any, None]:
dbos_sqlite_file = tmp_path_factory.mktemp('dbos') / 'dbostest.sqlite'
dbos_config: DBOSConfig = {
'name': 'pydantic_harness_dbos_tests',
'name': 'pydantic_ai_harness_dbos_tests',
'system_database_url': f'sqlite:///{dbos_sqlite_file}',
'run_admin_server': False,
'enable_otlp': False,
tests/_code_mode/test_temporal.py
+2 -2
View File
@@ -38,12 +38,12 @@ from pydantic_ai.messages import ModelRequest, ModelResponse, TextPart, ToolCall
from pydantic_ai.models.function import AgentInfo, FunctionModel
from pydantic_ai.toolsets.function import FunctionToolset
from pydantic_harness import CodeMode
from pydantic_ai_harness import CodeMode
pytestmark = pytest.mark.anyio
TEMPORAL_PORT = 7244 # avoid conflict with other test suites
TASK_QUEUE = 'pydantic-harness-code-mode-queue'
TASK_QUEUE = 'pydantic-ai-harness-code-mode-queue'
BASE_ACTIVITY_CONFIG = ActivityConfig(
start_to_close_timeout=timedelta(seconds=60),
retry_policy=RetryPolicy(maximum_attempts=1),
tests/test_placeholder.py
+3 -3
View File
@@ -3,12 +3,12 @@ from pathlib import Path
from pydantic_ai import Agent
from pydantic_ai.models.test import TestModel
import pydantic_harness
import pydantic_ai_harness
def test_import():
assert pydantic_harness.__doc__ is not None
assert isinstance(pydantic_harness.__all__, list)
assert pydantic_ai_harness.__doc__ is not None
assert isinstance(pydantic_ai_harness.__all__, list)
def test_test_model_fixture(test_model: TestModel):
uv.lock
Generated
+55 -55
View File
@@ -863,6 +863,61 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/5a/87/b70ad306ebb6f9b585f114d0ac2137d792b48be34d732d60e597c2f8465a/pydantic-2.12.5-py3-none-any.whl", hash = "sha256:e561593fccf61e8a20fc46dfc2dfe075b8be7d0188df33f221ad1f0139180f9d", size = 463580, upload-time = "2025-11-26T15:11:44.605Z" },
]
[[package]]
name = "pydantic-ai-harness"
source = { editable = "." }
dependencies = [
{ name = "pydantic-ai-slim" },
]
[package.optional-dependencies]
code-mode = [
{ name = "pydantic-monty" },
]
dbos = [
{ name = "pydantic-ai-slim", extra = ["dbos"] },
]
temporal = [
{ name = "pydantic-ai-slim", extra = ["temporal"] },
]
[package.dev-dependencies]
dev = [
{ name = "anyio", extra = ["trio"] },
{ name = "coverage" },
{ name = "logfire", extra = ["httpx"] },
{ name = "pydantic-ai-harness", extra = ["code-mode"] },
{ name = "pytest" },
{ name = "pytest-anyio" },
]
lint = [
{ name = "pyright" },
{ name = "ruff" },
]
[package.metadata]
requires-dist = [
{ name = "pydantic-ai-slim", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-ai-slim", extras = ["dbos"], marker = "extra == 'dbos'", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-ai-slim", extras = ["temporal"], marker = "extra == 'temporal'", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-monty", marker = "extra == 'code-mode'", specifier = ">=0.0.10" },
]
provides-extras = ["code-mode", "dbos", "temporal"]
[package.metadata.requires-dev]
dev = [
{ name = "anyio", extras = ["trio"] },
{ name = "coverage" },
{ name = "logfire", extras = ["httpx"], specifier = ">=4.31.0" },
{ name = "pydantic-ai-harness", extras = ["code-mode"] },
{ name = "pytest" },
{ name = "pytest-anyio" },
]
lint = [
{ name = "pyright", specifier = ">=1.1.408" },
{ name = "ruff", specifier = ">=0.14" },
]
[[package]]
name = "pydantic-ai-slim"
version = "1.80.0"
@@ -1015,61 +1070,6 @@ dependencies = [
{ name = "typing-inspection" },
]
[[package]]
name = "pydantic-harness"
source = { editable = "." }
dependencies = [
{ name = "pydantic-ai-slim" },
]
[package.optional-dependencies]
code-mode = [
{ name = "pydantic-monty" },
]
dbos = [
{ name = "pydantic-ai-slim", extra = ["dbos"] },
]
temporal = [
{ name = "pydantic-ai-slim", extra = ["temporal"] },
]
[package.dev-dependencies]
dev = [
{ name = "anyio", extra = ["trio"] },
{ name = "coverage" },
{ name = "logfire", extra = ["httpx"] },
{ name = "pydantic-harness", extra = ["code-mode"] },
{ name = "pytest" },
{ name = "pytest-anyio" },
]
lint = [
{ name = "pyright" },
{ name = "ruff" },
]
[package.metadata]
requires-dist = [
{ name = "pydantic-ai-slim", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-ai-slim", extras = ["dbos"], marker = "extra == 'dbos'", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-ai-slim", extras = ["temporal"], marker = "extra == 'temporal'", git = "https://github.com/pydantic/pydantic-ai.git?subdirectory=pydantic_ai_slim&branch=main" },
{ name = "pydantic-monty", marker = "extra == 'code-mode'", specifier = ">=0.0.10" },
]
provides-extras = ["code-mode", "dbos", "temporal"]
[package.metadata.requires-dev]
dev = [
{ name = "anyio", extras = ["trio"] },
{ name = "coverage" },
{ name = "logfire", extras = ["httpx"], specifier = ">=4.31.0" },
{ name = "pydantic-harness", extras = ["code-mode"] },
{ name = "pytest" },
{ name = "pytest-anyio" },
]
lint = [
{ name = "pyright", specifier = ">=1.1.408" },
{ name = "ruff", specifier = ">=0.14" },
]
[[package]]
name = "pydantic-monty"
version = "0.0.11"