DougTrajano/pydantic-ai-skills

Registry 信息

项目简介

pydantic-ai-skills 将 Agent Skills（agentskills.io）集成到 Pydantic AI 中，支持渐进式披露，并提供文件系统和程序化技能定义，用于构建模块化 AI 代理。

要点

• 代理技能的渐进式披露
• 基于文件系统的技能定义
• 程序化技能定义
• 与 Pydantic AI 无缝集成
• 兼容 Claude Code 和代理工作流

使用场景

• 使用可复用技能构建模块化 AI 代理
• 通过文件系统或代码管理技能库
• 为 Pydantic AI 代理添加结构化技能支持
• 使用 Claude Code 开发代理型 AI 应用

README 摘要

# pydantic-ai-skills A standardized, composable framework for building and managing Agent Skills within the Pydantic AI ecosystem. **Agent Skills** are modular collections of instructions, scripts, tools, and resources that enable AI agents to progressively discover, load, and execute specialized capabilities for domain-specific tasks. This package implements the [Agent Skills specification](https://agentskills.io/home) for Pydantic AI using a tool-calling approach, maintaining full compatibility with the specification while adapting the loading mechanism for Pydantic AI's architecture. Agent Skills is an open format maintained by Anthropic and open to contributions from the community. [](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://sonarcloud.io/summary/new_code?id=DougTrajano_pydantic-ai-skills) ## Features - **Progressive Disclosure** - Load skill information only when needed, reducing token usage - **Agent Skills Spec** - Fully compatible with the [Agent Skills](https://agentskills.io/home) open format - **Filesystem Skills** - Define skills as filesystem directories with Markdown files and scripts - **Programmatic Skills** - Create skills dynamically in Python code using decorators or dataclasses - **Type-Safe** - Built with Python dataclasses and type hints - **Validation** - Automatic validation of skill metadata and structure - **Multiple Directories** - Load skills from multiple sources - **Security** - Path traversal prevention and safe script execution

话题