Create, manage, and execute reusable code-based tool compositions for AI agents.
agent-skills is worth checking the docs before setup with trust notes worth reviewing. Check agent compatibility and use-case fit before adding it to your workflow.
npx skills add datalayer/agent-skillsRun the command in your terminal.
Confirm that the skill files were added to your agent workspace.
Check the README requirements before invoking the skill in your agent.
Agent Skills lets AI agents build their own toolbox by writing Python files. Agents can discover, create, and reuse skills to perform tasks. It works with Pydantic AI and other frameworks.
Agent Skills is a Python library that enables AI agents to create, manage, and execute reusable code-based tool compositions. Skills are Python files in a 'skills/' directory with async functions. Agents can discover skills by listing the directory, create new skills by writing files, execute them by importing and calling, and compose multiple skills together. The library supports both path-based loading (scanning directories for SKILL.md files) and module-based loading (importing skills from installed packages). It integrates with Pydantic AI via AgentSkillsToolset and includes a sandboxed executor for safe execution. The project is open-source under BSD-3-Clause license and is maintained by Datalayer.
Looks usable, but maintenance, license, or security notes deserve a closer look.
Last commit was about 12 days ago.
9 GitHub stars indicate community interest.
1 open issues signal maintenance load.
BSD-3-Clause license detected.
Enable AI agents to dynamically create and use custom tools during a conversation
Build a library of reusable skills for web crawling, PDF processing, GitHub automation, and event handling
Allow agents to compose multiple skills together to solve complex tasks
Distribute skills as part of a pip-installable package for team-wide reuse
Sandbox execution of agent-created code to prevent security risks
Agent skills can change coding-agent behavior; review every SKILL.md and referenced script before installing.
Check shell, network, file-system, credential, and API-key requirements before running skill workflows.
Executing arbitrary code from agents can be dangerous; always use a sandboxed executor
Skills may access filesystem or network; restrict permissions in sandbox
Malicious skill files could compromise the host if not sandboxed
9
Stars
1
Forks
1
Issues
BSD-3-Clause
License
A public repository of example skills for Claude, demonstrating how to create reusable instructions and resources for specialized tasks.
Adversarial AI bug hunter with auto-fix skill for Claude Code, Cursor, Codex CLI, GitHub Copilot CLI, Kiro CLI, Opencode, Pi Coding Agent, and more. Multi-agent pipeline finds security vulnerabilities, logic errors, and runtime bugs — then fixes them autonomously on a safe branch.
5 security/trust notes recorded.
Setup difficulty is 3/5.
An open-source framework to turn HTML, CSS, and animations into deterministic MP4 videos.