AI-agent readiness, scored.
An offline CLI that grades any repository 0–100 on how safely an AI coding
agent can work in it — deterministic, no telemetry, a concrete fix for every gap.
Website · Docs · Source · Marketplace
brew install use-charter/tap/charter
charter doctorAI agents are only as good as the repo they land in — missing context, exposed secrets, and unpinned tools quietly wreck their output. Charter catches that before an agent does: 18 rules across 9 categories (context, secrets, MCP safety, agent config, environment, CI, testing, autonomy, governance), a per-category scorecard, diff-first auto-fix, and SARIF straight into GitHub Code Scanning. It runs in under two seconds, makes zero network calls, and never sends your code anywhere.
- charter — the CLI, rules engine, and GitHub Action. Go · Apache-2.0 · SLSA Level 3.
- charter-action — gate every pull request on a readiness threshold.
Free and open source, forever · no LLM in the core · the score only ratchets up.