About RepoFinder

Ask for any kind of GitHub repository in plain English. Get ranked, scored, AI-explained results — grounded only in real GitHub data.

What is RepoFinder?

RepoFinder is a developer tool that lets you search GitHub repositories using plain English instead of GitHub's query syntax. Instead of figuring out the right language:rust stars:>1000 incantation, you just describe what you want.

The system has two layers: an AI compiler that translates your question into a precise GitHub search query, and a scoring engine (RepoRadar) that ranks the results by quality across five dimensions.

Everything runs in your browser. There is no database, no user accounts, and no server-side storage of your searches. The only server-side component is a lightweight proxy that calls an AI provider on your behalf — so your API key stays private.

How grounding works

"Grounding" means the AI is only allowed to discuss repositories that were actually returned by the GitHub API for your search. It cannot reference any repository that wasn't in the live results set.

This prevents the model from hallucinating popular-sounding project names. Every repository the AI mentions in its verdict must match — by exact full_name — a real record we fetched. Any mismatches are silently dropped before the verdict is displayed to you.

If GitHub returns zero results for a query, we skip the AI verdict entirely and show you a plain "no results" message.

The RepoRadar Score

Each repository is scored 0–100 across five dimensions:

Popularity (35%)Stars and forks, normalized on a log scale.

Freshness (25%)Days since the last push. Decays exponentially after 30 days.

Maintenance (20%)Open issue ratio and whether the repo is archived.

Community (10%)Fork-to-star ratio and organizational ownership.

Completeness (10%)Presence of description, license, topics, and homepage.

Tiers: Elite (90+), Strong (75–89), Solid (60–74), and Mixed (<60). Hover any score gauge for the five-dimension breakdown.

What's stored — and where

Location

What

Your browser (localStorage)

Recent searches (last 8), display preferences

Your browser (sessionStorage)

Last search results, cached for 10 minutes to avoid re-fetching

Our AI proxy server (transient)

Your question text + public GitHub metadata — used to call the AI, never persisted

GitHub's API (third-party)

Your search query is sent to GitHub search. See github.com/privacy

Use the "Clear my data" button on the home page to wipe all locally-stored data instantly.

No login. No database.

RepoFinder has no user accounts, no sign-up flow, and no server-side database. Search history lives only in your browser. Closing a private/incognito window removes everything.

GitHub search calls are made directly from your browser to GitHub's public API — no personal authentication token is required or used.

GitHub disclaimer

RepoFinder is an independent tool and is not affiliated with, endorsed by, or sponsored by GitHub, Inc. GitHub is a trademark of its respective owner. We use the name "GitHub" only in a descriptive/nominative sense to indicate the data source.

Contact

Questions, feedback, or concerns? Email us at hello@repofinder.xyz.