Can it find work that hasn't been merged into the main branch?

Yes, its primary methodology is branch-level divergence analysis. Testing shows that over 50% of substantive community work lives exclusively on feature branches rather than the fork's default branch.

Why does this skill ignore star counts when evaluating forks?

Empirical data shows that star counts are often anti-correlated with technical value. Many high-value forks are created for specific institutional or production needs and never promoted socially, making them invisible to star-based filtering.

How does this skill handle the 'noise' of thousands of inactive forks?

It uses timestamp clustering to identify forks that were automatically created but never modified. By skipping forks with 'pushed_at' timestamps identical to the upstream, it eliminates roughly 85% of inactive noise immediately.

What are the strongest signals of a 'Tier 1' fork?

The strongest signals include independent release tags, high 'ahead_by' counts on feature branches, and commit email domains belonging to major institutions or specialized companies.

Fork Intelligence

Name: Fork Intelligence
Author: terrylica

byterrylica

•

開発者ツール

Discovers high-value code in GitHub fork ecosystems by analyzing branch-level divergence and contributor signals beyond simple star counts.

Fork Intelligence provides a systematic framework for Claude to discover substantive work hidden within GitHub forks that traditional filtering methods often miss. By moving beyond star counts to analyze branch-level divergence, timestamp clustering, and institutional contributor signals, this skill allows developers to surface unique features, bug fixes, and architectural improvements from the broader ecosystem. It is an essential tool for maintainers looking to cherry-pick valuable community contributions or researchers evaluating the true state of an open-source project's evolution.

主な機能

01Branch-level divergence analysis to find features hidden on non-default branches

0220 GitHub stars

03Domain-specific heuristics for Scientific, Finance, and Infrastructure repositories

04Timestamp clustering to automatically filter out up to 85% of noisy bulk mirrors

05Cross-referencing upstream PR history to recover rebased or force-pushed work

06Signal-based ranking using commit email domains and independent release tags

ユースケース

01Auditing an ecosystem to identify institutional contributors and professional-grade extensions

02Discovering production-ready features in forks of abandoned or slow-moving repositories

03Systematic cherry-picking of community improvements for local repository enhancement

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add terrylica/cc-skills fork-intelligence

For use in Claude.ai and ChatGPT

Download Skill