Skip to content

🧩 Agent Skills & Libraries — Behavior-Based Ranking

Updated: 2026-08-10

How to read this table

ColumnMeaning
Used-ByCode-search manifest references: "<name>" filename:package.json OR filename:requirements.txt OR filename:pyproject.toml
StarsstargazerCount
Δ StarsChange since previous snapshot
Score0.35·log(1+deps) + 0.25·log(1+stars) + 0.2·log(1+Δstars) + 0.2·log(1+forks)

Top 30

#RepoUsed-ByStarsForksΔ StarsScore
1langchain-ai/langchain342,528143,83123,961+24310.546
2langchain-ai/langgraph108,28839,3276,606+2499.566
3microsoft/autogen66,43260,3379,090+549.263
4crewAIInc/crewAI16,83256,8788,109+1568.954
5agno-agi/agno53,50441,6385,749+278.868
6run-llama/llama_index10,32051,5167,898+778.613
7modelcontextprotocol/python-sdk12,12823,9543,764+338.164
8pydantic/pydantic-ai7,42419,1812,496+697.999
9mastra-ai/mastra5,90427,0622,603+587.979
10langgenius/dify78151,89723,977+2657.646
11vercel/ai1,10626,1034,934+457.462
12modelcontextprotocol/typescript-sdk2,37213,1082,064+197.216
13openai/openai-agents-python11028,5164,473+596.713
14anthropics/skills1167,26619,957+4946.471
15jina-ai/serve021,8622,24104.041
16cohere-ai/cohere-toolkit03,17842703.228

← Home · Methodology · MCP Servers

MIT License — data is free to reuse