✦About cyrus.ai

Building the
honest index for
the agent era.

Contributors today juggle a dozen feeds to find one issue worth taking, one org that reviews, or one program window still open. We exist to make finding your next contribution simpler - and every number behind that decision verifiable.

Verified statsOfficial sourcesLive open issuesAccepted-project archiveFree forever

Our mission Start exploring

cyrus.ai google

✦One contributor platform

Everything you need. One index.

Click any capability to see what's inside.

Skill catalog1528 agent skills, Cursor rules, AGENTS.md configs and MCP servers across 6 shelves - ranked by live GitHub stats, never by hype.

Every shelf is filtered by what the repo itself declares - SKILL.md, rules files, MCP manifests - and each card carries stars, forks, last-push and license from the public API.

Open it

Open issues, counted40k open issues in 290 of the 9k catalog repositories, crawled 2026-10-21.

An issue is only listed when GitHub still reports it open. Age, labels and the repository's own star count travel with it, so a stale invitation is obvious before you spend an evening on it.

Open it

Organization profilesContributor pulse before you commit: language mix, open issues, review latency, past accepted projects and who actually mentors there.

Profiles join our live GitHub directory with the accepted-project archive, so a page shows both the code signal and the program history for the same organization.

Open it

Program calendarGSoC, Outreachy, LFX, ESOC, MLH Fellowship, GSSoC, KDE, Summer of Bitcoin, Nexus Spring of Code, 24 Pull Requests and Hacktoberfest - 13 tracked windows on one timeline.

Each program page shows the phase you are actually in today - organization applications, participant applications, community bonding, coding period - with the window dates the organizer publishes.

Open it

Project archive8.8k program projects from the 12 archives we can read, 2017 to 2026, filterable by year, organization, stack and difficulty.

Pulled from each program's public API by our refresh pipeline - titles, orgs, mentors, tech tags and descriptions, never hand-typed.

Open it

Your boardIssues you are tracking through saved, interested, working and done; the repositories you follow; the languages that come out of both.

Sign-in is demo-grade on purpose: everything lives in your browser's localStorage until a session backend is attached. No email, no tracking pixel, no spam.

Open it

Why we exist

Discovery took longer than shipping.

The work moved: agents, skills, programs that expect a real merged pull request before the application window closes. The discovery layer didn't. Finding a beginner-labelled issue an org still has open, an org that reviews PRs in under three days, or a program deadline you can still hit means twelve tabs, three Discords and a gut feeling.

So we built the index we kept wishing for: 40k open issues crawled from 290 repositories on 2026-10-21, a skill catalog of 1528 repositories across 6 shelves published by 364 hand-picked organizations, 272 organizations in the issue crawl, the 13 mentored programs they join every year, and 8.8k program projects from 12archives, 2017 to 2026 - each with the boring-but-decisive facts: stars, forks, open issues, language, license and last push.

The directory is opinionated on purpose. A 67-skill pack by a solo builder sits next to Anthropic's official registry; an Outreachy internship is listed beside Google Summer of Code. If developers star it and it behaves like a skill, it's on a shelf.

Our principles

Guided by four rules.

1Verify, don't vibe.

Every number starts from a public API snapshot regenerated by a script - stars, forks, open issues, license, last push. Nothing is typed by hand, and estimates are labelled as estimates.

2One page over twelve tabs.

An issue carries its labels, its age and the repository it belongs to; an org carries its contributors, review latency and accepted projects. Both answer the same question - can I start this weekend - without a second tab.

3Credit where it's due.

Official org logos come from GitHub itself, every issue and pull request links to the real thread on github.com, and program pages name the organizer rather than paraphrase them.

4Free, and visibly so.

No account to browse, no email to save, no dark patterns. Demo sign-in keeps everything in your browser's localStorage, and the whole index stays open.

Data & methodology

Static stats come from a GitHub API snapshot dated 2026-10-02, regenerated by a script that walks the seed list, refetches each repo and rewrites the dataset - nothing is typed by hand. Repo and organization pages then re-fetch contributors and language mixtures live, and fall back to clearly-labelled deterministic estimates when the API is rate-limited. The issue index is a separate crawl of GitHub's search API, last run 2026-10-21, and the accepted-project archive refreshes from each program's public API on the same cadence.

Ranking - stars first; forks and last-push always shown so ranking context survives.
Difficulty labels - derived, not claimed: beginner under 10k★, intermediate to 50k★, advanced above.
Beginner signal - an issue counts as beginner-labelled only when GitHub carries a real good-first-issue or help-wanted label: 757 of 40,017 right now.
Logos - official org avatars via GitHub, and program covers are our own generated SVG artwork, never copied branding.
Sign-in - Google and email codes start a real session: an httpOnly cookie this server signs, not a value the page writes for itself. Your browsing state and saved items stay in your browser until a database is attached to the deploy.

Frequently asked

Where does the data come from?

Every number on cyrus.ai starts from the public GitHub API. We snapshot the catalog once a fortnight - stars, forks, open issues, license, last push - and repo pages re-fetch contributors and language breakdowns live when you open them.

How do you decide what counts as a “skill” or an “MCP server”?

A repo lands on a shelf if its own README says so: SKILL.md files for Agent Skills, .cursor/rules or mdc files for Cursor rules, AGENTS.md / DESIGN.md for configs, and an mcp server manifest or registry entry for MCP servers. Ambiguous repos go to the Dev Agents shelf.

Why rank by stars at all, if stars are hype?

They're the only cross-repo signal that's public, historical and cheap to verify. We show forks, open issues and last-push beside every star count so a 200k repo that hasn't moved in a year doesn't pretend otherwise.

How current are the open issues?

The issue index is a real crawl of GitHub's search API, last run 2026-10-21 across 290 repositories. An issue is listed only while GitHub still reports it open, but a repo can merge or close one between crawls - so every card links to the live thread and says how long ago it was touched.

Can I list my own repository?

Yes - save a bookmark on /projects for the submit flow we're shipping next: paste a GitHub URL, we enrich the metadata from the API, and your repo joins the review queue for the right shelf.

Is this affiliated with GitHub, Google or Anthropic?

No. Logos belong to their owners, event names belong to their organizers, and this is an independent index. If anything here mislabels your project, tell us and we'll fix the snapshot.

A good index is earned one commit at a time.

Snapshots refresh every fortnight, dates get re-checked against organizers, and corrections from orgs and builders ship the same week. That's the whole trick - and the whole promise.

Browse the catalog Open your dashboard

✦The companies behind the index

The data comes from teams that ship at scale.

Every org profile, star count and program record on cyrus is pulled from the public GitHub footprint of organizations like these - then re-verified on a schedule.

GitHub

The Apache Software Foundation

Browse all organizations