# Reporails

> Diagnose and govern AI instruction systems for coding agents — 92+ deterministic rules across Claude, Codex, Copilot, Cursor, Gemini.

Reporails diagnoses and governs AI instruction systems — the rules, system prompts, agent descriptions, and hooks that steer LLM coding agents (Claude, Codex, Copilot, Cursor, Gemini). It runs 92+ deterministic rules locally, scores instruction quality on a 0-10 scale, and provides actionable fixes.

## Quick start

Run a one-off check (no install, no account):

- npx @reporails/cli check
- uvx --from reporails-cli ails check

## Install permanently

Puts `ails` on your PATH:

- npx @reporails/cli install
- uvx --from reporails-cli ails install

## Key concepts

- **Rules**: 92+ deterministic checks across 6 categories (Structure, Coherence, Direction, Efficiency, Maintenance, Governance)
- **Server diagnostics**: Cross-file conflict detection, reinforcement analysis, per-instruction compliance scoring (sign in with `ails auth login` for full detail)
- **Supported Agents**: Claude, Codex, Copilot, Gemini, Cursor ? auto-detected from project files

## Links

- Rules reference: https://reporails.com/rules
- CLI source: https://github.com/reporails/cli

## Articles

Long-form notes on instruction quality, agent diagnostics, and what we measured in the 30k-corpus.

- Why your CLAUDE.md ignores half its instructions: https://reporails.com/articles/why-your-claudemd-ignores-half-its-instructions