# Izri

> AI QA for AI-generated code. When a coding agent (Claude Code, Cursor, Aider, Devin, your own bot) writes a PR, Izri answers one question: is this delta safe to merge? It does that by computing four independent signals — scope, tests, hallucination, visual — and aggregating them into a single `izri/quality` umbrella verdict.

Izri is a quality gate, not a code-review tool. Findings are addressed to the human — we never propose patches.

## Getting started

- [Integrations overview](https://izri.dev/integrations): the three install paths (GitHub Action, CLI, MCP) with copy-pasteable snippets.
- [Pricing](https://izri.dev/pricing): Free / $9 / $29 / $99 per organization, flat — plans, run quotas, what each tier unlocks. Also at [docs/pricing.md](https://izri.dev/docs/pricing.md).
- [Quick reference](https://izri.dev/docs/QUICK_REFERENCE.md): the one-page cheat sheet for first contact.
- [Installation & local dev](https://izri.dev/getting-started): clone, env, services, first run.
- [CLI](https://izri.dev/cli): `izri run`, `izri check-scope`, `izri status` — terminal surface for the four signals.
- [GitHub Action setup](https://izri.dev/github-action): drop-in workflow file, secrets, branch-protection wiring.
- [MCP server](https://izri.dev/mcp): mount `@izri/mcp` in Claude Code / Cursor / Windsurf for in-agent scoring tools.

## Four signals

- [Features overview](https://izri.dev/features): the four signals + the izri/quality umbrella verdict.
- [`izri/scope`](https://izri.dev/features/scope): does the diff match the PR's stated intent? Does it touch sensitive paths it shouldn't?
- [`izri/tests`](https://izri.dev/features/tests): did the relevant tests run and pass against the diff?
- [`izri/hallucination`](https://izri.dev/features/hallucination): do the tests that pass actually exercise the lines that changed?
- [`izri/visual`](https://izri.dev/features/visual): visual regression on changed routes. Premium-tier.
- [`izri/quality` umbrella](https://izri.dev/docs/features/scoped-test-selection.md): worst-of hard-rules + composite soft-signal aggregation across the four children above.

## Reference

- [Architecture](https://izri.dev/docs/architecture/README.md): system design, data flow, surface adapters.
- [Backend](https://izri.dev/docs/backend/README.md): Hono + tRPC API, analyzers, webhook receivers.
- [Frontend](https://izri.dev/docs/frontend/README.md): React Router 7 SSR dashboard.
- [Database](https://izri.dev/docs/packages/database-package.md): Drizzle + Postgres schema, migrations.
- [API reference](https://izri.dev/docs/api-reference/README.md): REST + tRPC procedures.
- [Webhooks](https://izri.dev/docs/features/webhooks.md): inbound webhook receivers and the GitHub App contract.
- [Outbound webhooks](https://izri.dev/docs/features/outbound-webhooks.md): firehose for delta-state events.

## Optional

- [llms-full.txt](https://izri.dev/llms-full.txt): the entire docs corpus concatenated. Use when you need everything in one fetch.
- [Repository](https://github.com/legendify-dev/izri): source code (open-core under AGPL-3.0).
- [Troubleshooting](https://izri.dev/docs/troubleshooting.md): common failure modes and their fixes.
- [Contributing](https://izri.dev/docs/contributing/README.md): how to ship a PR through Izri's own quality gate.

Every page on https://izri.dev/docs/ is also reachable as raw markdown by appending `.md` (e.g. `https://izri.dev/docs/cli.md`) or by sending `Accept: text/markdown`.