Leon-Drq/openagentskill

The skill layer for AI agents: npm for AI Agent Skills.

What it solves

OpenAgentSkill provides a decision layer for AI agents to find, evaluate, and install reusable "Agent Skills" (third-party code/tools) safely and effectively. It solves the problem of agents blindly installing popular but unsuitable or risky third-party repositories when trying to perform a specific task.

How it works

The project implements a task-to-skill resolver. When an agent provides a task, the system retrieves candidate skills, ranks them based on task fit, trust scores (incorporating maintenance, license, and risk signals), and real-world agent outcomes. It then provides a recommended skill, alternatives, and a stable install receipt for the agent to execute in a sandbox.

Who it’s for

  • AI Agent Developers: Those building agents (like Codex, Claude Code, or Cursor) that need a reliable way to extend their capabilities via external tools.
  • Skill Authors: Developers creating reusable agent tools that want a canonical page, trust badges, and a way to reach qualified users.

Highlights

  • Task-to-Skill Resolver: An API and CLI that converts a natural language task into a specific, reviewed install plan.
  • Trust Score v5: A ranking system that uses repository evidence and agent outcome feedback to signal risk and quality.
  • Agent Integration Kit: Ready-to-use templates for integrating the resolver into popular AI coding agents.
  • Machine-Readable Surfaces: Provides data via JSON, OpenAPI, and llms.txt for seamless agent consumption.
  • Outcome Loop: Allows agents to report whether a skill installation and run were successful or failed, improving future recommendations.

Related

  • Project
    aiskillstore/marketplace

    An open-source marketplace repository for AI agent skills, providing a secure distribution system and CLI for adding capabilities to tools like Claude Code and Codex.

  • Project
    HKUDS/OpenSpace

    A skill management layer for AI agents that enables them to retrieve, evaluate, and evolve their capabilities based on real-world task outcomes.

  • Project
    sickn33/agentic-awesome-skills

    A library of over 2,600 installable playbooks that allow AI agents to search, select, and apply reusable instructions for complex technical tasks.

  • Project
    kangarooking/kangarooking-skills

    A collection of structured, open-source Agent Skills compatible with the Agent Skills standard, providing reusable workflows for content creation, image generation, and AI project engineering.

  • Project
    zjunlp/SkillNet

    SkillNet is an open‑source platform that lets you discover, generate, evaluate, analyse and route reusable AI‑agent skills. It offers a public catalogue (500 K+ GitHub‑hosted skills), a Python SDK/CLI with six operations (search, download, create, evaluate, analyze, route), a local web UI, and integration points for agents via an MCP server. The system builds a graph of skill relationships and uses LLM‑powered routing (Claude/Codex) to pick the best skills for a task, providing safety/maintainability scores and reasoning.