1 / 3
【Overview】 Agent Skills Hub is an open-source directory of 250,000+ AI agent skills, MCP servers and agent tools, refreshed from GitHub every 8 hours. Its job is to answer a question that star counts cannot: is this skill safe to install before you run a stranger's code inside your agent? Every popular entry that has a fetched README is security-graded as Safe, Caution, Unsafe or Reject against 42 published rules built on the SlowMist agent-security taxonomy, and each tool also receives a 0–100 quality score covering completeness, clarity, specificity, examples, README structure and agent readiness. The result is a directory you can filter by more than popularity. It is aimed at developers building with AI agents, MCP servers and automation tools, and at teams that have to decide which open-source AI tools to adopt. Access is free. 【Key features】 • Security grading: Every skill, MCP server and agent tool with a fetched README is graded Safe, Caution, Unsafe or Reject against 42 published rules built on the SlowMist agent-security taxonomy. • Quality score: A 0–100 composite score across completeness, clarity, specificity, examples, README structure and agent readiness. • Skill Analyzer: Paste any public GitHub repo URL to get an instant security analysis with the matching rules, trust tier and quality score. • Side-by-side comparison: Line up several AI agent tools at once and read stars, quality and activity in a single view. • Scenario shortlists: Ranked, evidence-backed answers to "what's the best tool for X?" across categories such as browser automation, code review and database MCP servers. • Weekly trending digest: A weekly email covering the top 20 movers by star velocity, plus new MCP servers and top skill picks. 【Use cases】 1. Check whether a skill or MCP server is safe to install before running it. 2. Compare AI agent tools side by side before committing to one. 3. Find the best tools for a task such as browser automation, code review or database management. 【Pros and cons】 Pros: - Anyone about to run third-party code inside an agent gets a fast, published-rubric safety signal, which is far more actionable than a popularity ranking when the risk is execution rather than inconvenience. - Because the directory is open-source and free, teams weighing open-source AI tools can assemble a shortlist and a documented rationale without any procurement or signup step. - The regular refresh and weekly digest suit people who want to keep up with new MCP servers and skills through one stream instead of repeated manual scanning. Cons: - Security grading reads the README, not the code, so a Safe grade reflects what the repository documents rather than a review of what it actually executes. 【FAQ】 Q: What do the security grades mean? A: Entries are graded Safe, Caution, Unsafe or Reject against 42 published rules built on the SlowMist agent-security taxonomy, applied to the repository's fetched README. Q: How current is the directory? A: Listings are refreshed from GitHub every 8 hours, and a weekly email highlights the top 20 movers by star velocity along with new MCP servers and top skill picks. Q: Can I analyze a repository that is not already listed? A: Yes. The Skill Analyzer accepts any public GitHub repo URL and returns an instant analysis with the rules that applied, the trust tier and the quality score.
free
WPS AI is an AI assistant built into WPS Office that helps users write documents, generate presentation outlines and…
Baidu Comate is an AI coding assistant built by Baidu on its ERNIE model. It goes beyond simple autocomplete: it…
UniConverter is an AI media conversion and enhancement tool for video, audio, and images. It combines everyday…
6 similar tools to consider instead of Agent Skills Hub