Find the AI agents, skills & tools worth your time.

AgentSpot is a free, independent directory that helps you discover popular AI agents, reusable agent skills, and tools that agents can use. Every listing links out to the project's own repository and website — it is a hub for finding projects, not a platform that hosts or runs them. Listings are refreshed automatically from public data.

We rank primarily by GitHub stars, with package downloads (npm and PyPI, last 30 days) as a secondary signal where available. Each listing shows the raw numbers so you can judge for yourself, plus a 30-day growth figure once there is enough history. The composite popularity score is a transparent, log-scaled blend of these public metrics — and it is never influenced by payment.

Projects that meet our published thresholds (currently at least 100 GitHub stars, or 10,000 downloads in the last 30 days) are discovered and added automatically by a daily pipeline after review. You can also submit a project yourself; if it already meets the threshold it goes to review, and if not, the pipeline will list it automatically once it qualifies.

No. This is a hard rule: no payment of any kind — sponsorship, curation fee, or anything else — can affect whether a project is listed, how it is ordered, or its popularity score. Sponsored placements are always clearly labeled and shown in a separate slot, and the organic rankings never move. That independence is the whole point of the hub.

Not today. AgentSpot is a free directory: every listing links out to the project's own repository or website, and nothing is sold here. We have considered a vetted marketplace layer — independently security-tested listings with clear provenance — built on the same no-paid-rankings rules: if that ever arrives, it will never affect what is listed or how anything ranks.

Not necessarily. A listing is not an endorsement or a safety guarantee. Agents, skills, and tools execute code and can cause real harm, so treat everything here as a starting point for your own review. Our standing recommendation: read the source, pin the version, and ideally have your own agent rebuild the capability from the linked repository rather than installing published artifacts blindly.