About DiscoverGit

Automated open source intelligence — not a hand-picked list

DiscoverGit is not a hand-picked list. It is an automated discovery system for open source repositories.

No repository can buy placement. No repository is manually promoted. Visibility on DiscoverGit comes entirely from public signals, growth data, AI analysis, and a set of documented catalog rules.

The goal is to surface strong open source projects earlier, with less noise — without pretending that a human editor made those choices.

Who we are

DiscoverGit is an independent project based in Italy. It is built for developers, founders, maintainers, and technical teams who follow open source closely but do not have time to manually inspect hundreds of repositories every week.

We are not affiliated with GitHub. We use GitHub as the source of public repository metadata, then organize, filter, score, and explain that data to make discovery faster.

How the system works

DiscoverGit continuously indexes public GitHub repositories and computes signals automatically:

  • Stars, forks, open issues, license, language, topics, last activity
  • Weekly and daily star growth
  • Relative velocity compared with repository size and history
  • Anomaly detection when growth diverges from baseline
  • AI-generated summaries, quality scores, discovery value, tags, badges, and risk flags
  • Deep analysis reports for selected repositories

All of this is derived from public data. No private repository information is accessed. No owner or contributor is contacted.

There are three distinct stages in the catalog:

Imported

Repository exists in the database for tracking purposes. Not publicly visible.

Public catalog

Passes all visibility rules: active, not archived, not fossil, not excluded, not deleted. Appears in search, rankings, topics, and digest.

Deep analysis

Has a published, up-to-date, review-cleared AI report. A smaller subset of the public catalog.

How ranking works

The catalog has multiple sort modes. Each measures something different. Understanding the difference matters.

Trending

Default

Repositories with strong recent growth relative to their size. Combines star velocity, activity freshness, and anomaly signals. Not simply the most starred.

Best match

AI quality score weighted with discovery value. Avoids always surfacing the same giant projects: a well-documented famous project does not automatically outrank a smaller project that is actively maintained and clearly scoped.

Hidden Gems

Sorted by discovery value — how interesting a repository is to find before it becomes obvious. A repository known by everyone has low discovery value regardless of quality. This sort is designed to surface projects worth knowing earlier.

Fast movers

Sorted by anomaly score — repositories growing significantly faster than their own historical baseline. Anomaly does not mean quality. It means something changed: a viral post, a new release, an unexpected mention. DiscoverGit surfaces it; you decide if the reason matters.

Stars, Growth, Recent

Classic signals. Stars = cumulative popularity. Growth = stars gained in the last 7 days. Recent = newly indexed repositories, only shown once AI enrichment is available.

Caught early

Repositories DiscoverGit indexed before they took off — shown with a growth multiplier (e.g. 5×) from their star count at first indexing to today. A running track record of early picks.

How the AI works

DiscoverGit uses AI to read public repository metadata and README excerpts, then produce structured analysis.

The AI does not claim to inspect the full source code. When evidence is missing from public data, it is instructed to say so rather than guess. Analysis must pass an internal review check before it is published publicly.

Base enrichment runs on all catalog repositories and produces quality score, discovery value, activity signals, tags, and badges. Deep analysis is a longer report covering adoption evidence, mainstream potential, risks, and a recommendation on when a repository is worth adopting, monitoring, or avoiding.

What the scores are not

  • ✕ A guarantee of code quality or security
  • ✕ A substitute for reading the source code before adopting a dependency
  • ✕ A paid or sponsored ranking
  • ✕ An editorial judgment by a human curator
  • ✕ A promise that a project is actively maintained or suitable for production

The best use of DiscoverGit is not "let the system decide for me." The best use is "show me the signals worth a closer look, and explain why."

Why we built it

Open source discovery has two recurring problems.

The first is noise. Too many repositories, too many stale popularity signals, too many abandoned projects that still rank high from years-old stars.

The second is timing. By the time a project is obviously mainstream, the early discovery window is closed.

DiscoverGit is built on the idea that useful open source intelligence must combine quality, growth, freshness, discovery value, and context — not just stars. And that the rules behind that combination should be visible and documented.

Our long-term goal is to demonstrate that an automated signal-based system can consistently identify future mainstream projects months before they become obvious.

Transparency

When the ranking rules, catalog policy, or AI behavior changes, we document it. See the public changelog.

Contact

Questions, corrections, or feedback: staff@discovergit.app