pydantic-ai-harness

name	pydantic-ai-harness
description	Extend Pydantic AI agents with batteries-included capabilities from pydantic-ai-harness -- Code Mode (collapse many tool calls into one sandboxed Python execution), a filesystem and shell, sub-agents, planning, context compaction, and more. Use when the user mentions pydantic-ai-harness, CodeMode, Monty, code mode, or tool sandboxing, when they want first-party filesystem/shell/sub-agent/planning/compaction capabilities for a Pydantic AI agent, when they want an agent to run agent-written Python, or when a Pydantic AI agent would benefit from orchestrating multiple tool calls in a single sandboxed script.
license	MIT
compatibility	Requires Python 3.10+ and pydantic-ai-slim>=2.1.0
metadata	{"version":"0.1.0","author":"pydantic"}

Building with Pydantic AI Harness

Pydantic AI Harness is the official capability library for Pydantic AI. Capabilities that need model or framework support -- and those fundamental to every agent -- live in core pydantic-ai; optional, batteries-included capabilities live here. Both are composed onto an agent through the same capabilities=[...] API.

This skill covers the capabilities shipped by pydantic-ai-harness. For the core framework -- agents, tools, structured output, hooks, and testing -- use the building-pydantic-ai-agents skill instead.

When to Use This Skill

Invoke this skill when:

• The user mentions pydantic-ai-harness, CodeMode, code mode, or the Monty sandbox
• An agent makes many sequential tool calls that could collapse into one sandboxed Python execution
• The user wants the model to write Python that loops, branches, aggregates, or parallelizes tool calls with asyncio.gather
• The user asks to sandbox or constrain the code an agent runs

Do not use this skill for:

• Core Pydantic AI usage -- building agents, adding tools, structured output, streaming, or testing (use building-pydantic-ai-agents)
• Capabilities that ship in core pydantic-ai, such as web search, tool search, and thinking
• The Pydantic validation library on its own (pydantic/BaseModel without agents)

Supported Capabilities

CodeMode has a full reference below; it is the flagship capability and the one this skill goes deep on. The rest ship today and each has its own README with API and examples.

Stable (imported from pydantic_ai_harness):

Capability	Description	Reference
CodeMode	Wraps eligible tools into a single sandboxed run_code tool so the model orchestrates them in Python	Code Mode
FileSystem	Read, write, edit, and search files under a root directory, with traversal prevention	README
Shell	Run commands in a subprocess with allowlists, a default denylist, timeouts, and env masking	README
ManagedPrompt	Back an agent's instructions with a Logfire-managed prompt	README

Experimental (imported from pydantic_ai_harness.experimental, may change in any release): SubAgents, Planning, a compaction family (SlidingWindow, SummarizingCompaction, ...), OverflowingToolOutput, RepoContext, StepPersistence, PyaiDocs, RuntimeAuthoring, and an ACP server adapter.

The full, current list with links and status is in the capability matrix.

Install

uv add pydantic-ai-harness

Each capability declares its own extra. Code Mode needs the Monty sandbox:

uv add "pydantic-ai-harness[codemode]"   # `code-mode` is also accepted as an alias

Requires Python 3.10+ and pydantic-ai-slim>=2.1.0.

Quick Start

A harness capability is added to the agent like any other. Here CodeMode wraps an MCP server's tools into a single run_code tool that the model drives with Python.

from pydantic_ai import Agent
from pydantic_ai.capabilities import MCP  # MCP ships in core pydantic-ai

from pydantic_ai_harness import CodeMode

agent = Agent(
    'anthropic:claude-sonnet-4-6',
    capabilities=[
        # native=False routes the MCP tools through a local toolset so CodeMode can wrap them.
        # Without it, providers with native MCP run the tools server-side and bypass the sandbox.
        MCP('https://hn.caseyjhand.com/mcp', native=False),
        CodeMode(),
    ],
)

result = agent.run_sync(
    'Across the top and best Hacker News feeds, find the most-discussed story with at '
    'least 100 points and summarize its comment thread in one paragraph.'
)
print(result.output)
#> The most-discussed story clearing 100 points is ...

Instead of one model round-trip per tool call, the model writes a single Python script that fetches both feeds with asyncio.gather, dedupes and ranks them in plain Python, and pulls the winning thread -- collapsing many calls into one run_code.

Key Practices

• Confirm a harness capability is actually needed. If core Pydantic AI tools and capabilities are enough, use the building-pydantic-ai-agents skill instead -- don't reach for the harness by default.
• Read the reference before writing code. Each capability has its own configuration, constraints, and gotchas -- load the linked reference (e.g. Code Mode) first.
• Install the capability's extra. Importing CodeMode without pydantic-ai-harness[codemode] raises an ImportError; the Monty sandbox is an optional dependency.

Common Gotchas

• native=True tools bypass CodeMode. Provider-native MCP servers and web search execute server-side, so run_code never sees them. Construct them with native=False to keep them local and wrappable.
• The Monty sandbox is a Python subset. No class definitions, no third-party imports, and only a small stdlib allowlist -- read Code Mode before debugging generated code that fails to run.
• CodeMode needs its extra. Install pydantic-ai-harness[codemode], not the bare package.