About GitBored

How GitBored finds open source projects, tracks star growth, and suggests related repositories.

Repository discovery & submissions

GitBored indexes open source repositories from several automated feeds and community submissions:

Scheduled star sweeps: Workers periodically query GitHub star ranges (50k+, 10k–50k, and 100–10k across rotating languages).
Trending scans: Daily sweeps monitor active GitHub trending repositories.
Direct submissions: Anyone can submit a repository via the submit page or by replacing github.com with gitbored.com in any repository URL.
Renames & redirects: When a repository is renamed or moved on GitHub, scans follow the redirect and migrate history to the canonical name.

Submissions and updates run in the background. A busy queue can delay when new data appears.

Repositories are grouped by star count and recent commit activity:

QualifiedDaily rescans & snapshots

IncubatingWeekly checks

Active repositories that do not meet the qualified thresholds, plus projects that went quiet. Rechecked weekly and promoted when they qualify.

FilteredNot rescanned

Repositories under 100 stars with no push in over 6 months are excluded from scheduled scan queues.

Unlike static star counts, GitBored tracks star velocity over time to help highlight projects gaining momentum:

Daily snapshots: Starting at 03:00 UTC, workers record stars, forks, watchers, and open issues for qualified, non-archived repositories.
35-day window: Growth rates are computed over a rolling 35-day history window and stored on the project for fast sorting.

Growth numbers require accumulated history and are not available immediately upon indexing:

Daily growth: requires at least 2 distinct daily snapshots.
Weekly growth: requires at least 80% coverage (~6 days of snapshot history).
Monthly growth: requires at least 80% coverage (~24 days of snapshot history).

New repositories display "–" for growth periods that haven't collected enough snapshots yet.

Related projects use AI-generated summaries to compare what repositories do, rather than matching keywords alone:

Unified text profile: A structured summary is compiled from the repository name, description, language, topics, activity recency, codebase size, license, and key README sections (overview, usage, features).
AI comparison: Cloudflare Workers AI converts each summary into an embedding for similarity searches. Unchanged summaries do not need to be regenerated.
Balanced similarity ranking: Related repositories are ranked using 70% vector cosine similarity plus 30% star-tier proximity.

The star-tier weighting favors projects with similar popularity, alongside similarity in purpose.

Background maintenance runs on scheduled UTC intervals:

00:45 UTCHigh-churn discovery: Rotating language scan for repos between 100 and 10k stars.

01:00 UTCTrending sync: Daily scan of active GitHub trending repositories.

01:30 UTCQualified refresh: Updates stars and metadata for qualified projects not scanned in 24 hours.

03:00 UTCDaily snapshots: Captures metrics snapshot for all qualified, non-archived projects.

Sunday 02:30Incubating sync: Weekly scan of incubating repositories to check qualification progress.

Sunday 03:00Embedding refresh: Rebuilds semantic vectors for projects whose descriptions or READMEs changed.

You can verify real-time queue depths, worker status, and recent scan counts on the system health page.

Navigate GitBored quickly from any page using keyboard shortcuts:

GitBored is built by Boring Dystopia Development. Repository data is sourced directly via the official GitHub REST API.