Skyll

skyll

Skill discovery for AI agents

Give any AI agent the power to discover and use new skills on demand. Search, retrieve, inject into context.

Get Started

Skyll aggregates skills from multiple sources (like skills.sh, registries, and more), ranks them by relevance, and returns the best matches. Agents discover and learn skills autonomously.

MCP ServerRecommended

Hosted MCP

Add Skyll to any MCP-compatible client, or install as a skill for agents that support it. Either way, your agent can search, discover, and learn skills on demand.

Add to your MCP config:

{
  "mcpServers": {
    "skyll": {
      "url": "https://api.skyll.app/mcp"
    }
  }
}

Or add as a skill:

npx skills add assafelovic/skyll
Python

Client Library

Use the Python client for typed, async access. Perfect for custom agents where context engineering matters.

Install:

pip install skyll
from skyll import Skyll

async with Skyll() as client:
    skills = await client.search("react")
REST API

Direct API Access

Call the API from any language or framework. Fetch the latest version of any skill by name, or search across all sources with a single request.

Get a specific skill:

curl "https://api.skyll.app/skill/tavily-ai/skills/search"

Search for skills:

curl "https://api.skyll.app/search?q=testing&limit=5"

How Scoring Works

Skills are gathered in realtime from multiple sources (skills.sh, curated registry, and more), so results are always fresh. Each skill is ranked 0-100 based on: content availability (40 pts), query match (30 pts), popularity (15 pts), and references (15 pts).

Skills from the community-curated registry get a small boost (up to 8 pts) scaled by how relevant they are to your query. Think of it like a Wikipedia for skills - open for anyone to contribute, validated by the community. The boost helps surface trusted skills while preserving real popularity and relevance signals.

Semantic search with embeddings coming soon.

Read the Docs