pydantic-ai-skills

This package implements Agent Skills (https://agentskills.io) support with progressive disclosure for Pydantic AI. Supports filesystem and programmatic skills.

What is pydantic-ai-skills?

This package implements Agent Skills (https://agentskills.io) support with progressive disclosure for Pydantic AI. Supports filesystem and programmatic skills.

Overview of pydantic-ai-skills

What it is

pydantic-ai-skills is a Python package that implements Agent Skills support with progressive disclosure for Pydantic AI. It provides remote skill registries and bundled-file execution for Agent Skills in Pydantic AI. Agent Skills are modular packages of instructions, resources, and scripts that teach an agent to handle a specialized task. On disk, a skill is a folder containing a SKILL.md file with name, description, and Markdown instructions, plus optional reference documents and executable scripts. The package requires and delegates to pydantic-ai-harness for SKILL.md parsing, validation, catalog, and instruction rendering, and adds remote registries, bundled file tools, sandboxed execution, and programmatic skills.

Key features

  • Remote registries: load skills from Git repositories and S3 sources, with composition (filter, prefix, rename, merge).
  • Bundled files: read_skill_resource and run_skill_script tools allow a skill to read reference documents or run scripts as written.
  • Sandboxed execution: keep untrusted scripts off the host.
  • ${SKILL_DIR} resolution: substitutes the skill directory path in place of the placeholder left by harness.
  • Programmatic skills: define skills in Python using decorators or dataclasses, in the same catalog.
  • Skill selection: give each agent a subset of a shared library with include / exclude.

Use cases

  • Building a research assistant that can load and use skills from local directories or remote repositories.
  • Creating agents that need to access bundled reference documents or execute scripts within a skill.
  • Managing a large skill library with progressive disclosure so only the needed skill's instructions are loaded.
  • Running untrusted skill scripts in a sandboxed environment.

Getting started

Install the package:bashuv add pydantic-ai-skillsOr with pip:bashpip install pydantic-ai-skillsPoint a SkillsCapability at one or more skill libraries and add it to your agent:pythonfrom pydantic_ai import Agentfrom pydantic_ai_skills import SkillsCapabilityagent = Agent( model='gateway/openai:gpt-5.2', instructions='You are a helpful research assistant.', capabilities=[SkillsCapability('./skills')],)result = await agent.run('What are the last 3 papers on arXiv about machine learning?')print(result.output)Or pull skills from a Git repository:pythonfrom pydantic_ai_skills import GitSkillsRegistry, SkillsCapabilitycapability = SkillsCapability( './skills', registries=[GitSkillsRegistry('https://github.com/anthropics/skills', path='skills')],)

License and pricing

This package is open source under the MIT License. No pricing information is stated beyond the open source license.

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/dougtrajanopydantic-ai-skills.svg)](https://yunodb.com/tools/dougtrajanopydantic-ai-skills?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/dougtrajanopydantic-ai-skills?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/dougtrajanopydantic-ai-skills.svg" alt="Featured on YunoDB" height="20"></a>