Running the pre-flight inspection…
Running the pre-flight inspection…
Agent-Ready Score
70/100Flight-worthy, minor squawks
[](https://sortielabs.com/score/supabase/supabase)
Agent rules file7/15
Found .cursorrules, copilot-instructions.md but no AGENTS.md — only some tools will pick these up.
FIX → Add an AGENTS.md at the root: architecture map, hard rules, and how to verify changes.
Rules have substance0/8
No rules content to measure.
FIX → Write real working rules: what to never touch, what to run after every change, where things live.
Cross-tool coverage7/7
2 rule files — Claude Code, Cursor, and Copilot users all get guidance.
CI pipeline10/10
GitHub Actions workflows found — agents' mistakes get caught.
Tests exist8/8
Test files found — agents can prove their changes work.
Score another repo
One-command verify3/7
Lint + typecheck scripts exist, but there's no single verify command.
FIX → Add `scripts/verify.sh` (typecheck + lint + build + test) and require it in your agent rules.
Env template0/7
No .env.example found.
FIX → Ship a .env.example so agents (and humans) know what config exists without seeing values.
.gitignore5/5
Present.
Lockfile committed4/4
Dependencies are pinned — reproducible installs.
License4/4
Present.
README8/8
A real README exists.
Contributing guide4/4
Present.
Docs directory0/3
No docs/ directory.
FIX → A docs/ folder gives agents somewhere to look before inventing answers.
Typed foundations10/10
tsconfig.json found — the compiler catches what reviews miss.