RepoScore

Analyze GitHub repositories and generate a structured quality score.

What RepoScore Does

RepoScore evaluates repository quality from the signals developers actually use during adoption and review. It analyzes repository metadata, README quality, package metadata, and file-tree structure, then produces a score with actionable follow-up items.

The scoring pipeline covers:

repository analysis for any public GitHub repository
documentation scoring for onboarding quality
structure analysis for source layout and project organization
discoverability scoring for metadata, badges, topics, and docs links
agent safety scoring for repositories that expose AI runtime or autonomous-agent behavior

RepoScore powers the analysis engine behind RepoForge and is intended to be explainable. The same analysis pipeline is used for the API response, the public repository page, and the badge output.

Example Analysis

View the live RepoForge analysis page for RepoScore:

https://repoforge.dev/repos/repoforge-dev/repo-score

That page shows the current repository classification, overall score, category breakdown, and improvement suggestions generated from the live repository snapshot.

Installation

git clone https://github.com/repoforge-dev/repo-score.git
cd repo-score
npm install
npm test
node server.js

RepoScore runs on Node.js 18 or newer.

Usage

Analyze a repository through the public RepoForge API:

curl "https://repoforge.dev/api/analyze?repo=repoforge-dev/repo-score"

Example repository page:

curl "https://repoforge.dev/repos/repoforge-dev/repo-score"

Typical response fields:

repo: normalized owner/repo identifier
repoType: detected repository type
repoScore: overall weighted score
scores: category-level scores
improvements: concise repository improvements

How Scoring Works

RepoScore combines a fixed set of category analyzers:

documentation: README quality, installation guidance, examples, API/reference coverage, configuration notes, and contribution guidance
structure: source layout, tests, build configuration, formatting or lint configuration, and CI
discoverability: topics, badges, repository description, homepage metadata, and documentation links
maintenance: recent commits, releases, contributors, and visible issue activity
adoption: stars, forks, and contributor signals
agentSafety: safety documentation, guardrails, permission boundaries, and runtime limits for AI-related repositories

The final score is a weighted average based on the detected repository type. More detail is documented in docs/scoring.md and docs/api.md.

Badge

Add the RepoScore badge to a repository README:

[![RepoScore](https://repoforge.dev/badge/repoforge-dev/repo-score)](https://repoforge.dev/repos/repoforge-dev/repo-score)

Example integration:

[![RepoScore](https://repoforge.dev/badge/vercel/next.js)](https://repoforge.dev/repos/vercel/next.js)

Contributing

Contributions should keep scoring behavior deterministic and explainable. Update tests when analyzer rules or weighting behavior changes, and keep repository-quality improvement messages concise and actionable.

Start with CONTRIBUTING.md. Supporting documentation lives in docs/api.md, docs/scoring.md, and docs/architecture.md.

License

MIT

repoforge-dev/repo-score