Documentation 20%
README that explains the repo in 15 seconds, plus install command, usage example and a real changelog.
From FastPress Labs — page audited at A+ by our own LaunchGrade
Grade your repo before the internet does.
LaunchGrade audits your landing page. RepoGrade audits everything behind the link: 33 checks across docs, community, engineering, discoverability, release hygiene and agent-readiness — on any public GitHub repo, in one command. Zero dependencies. Free forever.
Reads public GitHub metadata only. Never uploads your code. Reports stay local unless you publish them.
A+96/100Announce-ready
Sample repo: fastpresspages/launchgrade · B 79 for fastpress-start · we publish low grades too.
What it grades
Six categories, weighted like a launch-quality scorecard. Warns cost half credit; passes cost none.
Community 15%
License (permissive preferred), contributing guide, code of conduct, support channels, issue and PR templates.
Engineering 25%
Tests that run, CI that runs them, declared toolchain, committed lockfile — and a gentle flag when dependencies pile up.
Discoverability 15%
Description, at least three topics, homepage link and current activity, so search and social cards actually work.
Release hygiene 10%
A tagged release, readable notes and a tag that matches package.json — where stability actually lives.
Agent-readiness 15%
AGENTS.md, an installable skill, an MCP or manifest, llms.txt — the layer humans forget and AI agents read first.
repograde --list
Install once, grade any repo
$ npx -p github:fastpresspages/repograde repograde fastpresspages/launchgrade
▲ RepoGrade v0.1.0 · fastpresspages/launchgrade
https://github.com/fastpresspages/launchgrade • branch main
┌──────────────────────────────────────────────┐
│ A+ 96/100 Announce-ready │
└──────────────────────────────────────────────┘
28 pass · 3 warn · 0 fail
Documentation 100 ████████████████████
Community 100 ████████████████████
Engineering 96 ███████████████████░
Discoverability 100 ████████████████████
Release hygiene 85 ████████████████░░░░
Agent-readiness 90 ██████████████████░░
Fix these first
1. FAIL [Agent] llms.txt published
No llms.txt at the repo root
2. WARN [Release] Tag matches package.json
package.json 0.3.0 vs release tag v0.1.0
→ Full report: samples/launchgrade/report.html (report.json/md, badge.svg, fix-plan.md)npx -p github:fastpresspages/repograde repograde owner/repo
-o report writes report.html, report.json, report.md, badge.svg and fix-plan.md. Use a GITHUB_TOKEN for 5,000 requests/hour. An npm package is planned once publishing credentials exist; the GitHub install path above works today.
Real artifacts from real scans
Questions people ask first
Is my repo private?
RepoGrade only reads public GitHub metadata via the API: root file tree, README, package.json, releases. It cannot see private content and it never writes to your repo. Reports are generated locally and stay wherever you save them.
What happens with repos without package.json?
Package-specific checks (test script, engines, zero-deps, lockfile, tag match) cleanly skip and don't count against you. Any repo is gradeable.
Can this run in CI?
Yes — a composite GitHub Action ships in the repo, or call npx -p github:fastpresspages/repograde repograde owner/repo --fail-under 80 in your workflow to gate releases on the score.
Where do the weights come from?
Engineering is heaviest (25%) because broken basics hurt adopters most; agent-readiness is 15% because that layer is invisible to humans and decisive for AI-helpers. Say pull in the repo if you'd like a different weighting proposal — the scoring is an honest heuristic, not a law.
Who built this?
FastPress Labs — small, free, open source tools for people who ship. LaunchGrade audits the landing page, RepoGrade audits the repository, and a free human review is always available when you want a person in the loop.
Free repo review (human actually reads it)
Drop your repo and what you're stuck on. We run RepoGrade and read the code/findings manually, then reply with the three changes that matter most. No fee, no obligation, no drip campaign.
Prefer plain mail? meetcode99@gmail.com. We reply to real messages from public repos only – send yours if the mailto is easier.