crit.md is built with Phoenix LiveView, hosted on Fly.io behind Cloudflare. It measures traffic with Umami.
Built with Phoenix LiveView, hosted on Fly.io behind Cloudflare.
| Technology | Version | First seen |
|---|---|---|
| JavaScript Frameworks & Libraries1 | ||
| – | 7 Nov 2026 | |
| Frontend Build Tools & CSS1 | ||
| – | 7 Nov 2026 | |
| Web Servers, Languages & Backend Frameworks2 | ||
| – | 7 Nov 2026 | |
| CDN & Hosting2 | ||
| – | 7 Nov 2026 | |
| Analytics & Tag Management1 | ||
| – | 7 Nov 2026 | |
| Accessibility1 | ||
| – | 7 Nov 2026 | |
| Error Tracking & Performance Monitoring1 | ||
| – | 7 Nov 2026 | |
| Video, Audio & Media Delivery1 | ||
| – | 7 Nov 2026 | |
The crew hasn’t read this site’s colors yet. We read a palette when we capture a site’s screenshot, busiest sites first. Meanwhile, browse palettes from other sites.
Counted from every stack change we've recorded: of the sites we watched running one of this site's tools, how many went on to add another. A pattern across sites, not a forecast for this one.
From what the site says about itself, and account IDs it shares with other sites.
mailserver.purelymail.comjeff.ns.cloudflare.comruth.ns.cloudflare.comgoogle-site-verification=tHbsHAN-EbU3vJ1nMzWzQ2NVOs3OzSXW_bZeMeB1-CYv=spf1 include:_spf.purelymail.com ~allpurelymail_ownership_proof=abc2a28c6346a27adf9fd9448ec2520952d6beea05064acbbe06fbdc3418f7b3f5ad9ad008bb8ff2c5950d9fca36f954aa6f9d587437832a2ce414ffbc7a319fv=DMARC1; p=reject; ruf=mailto:dmarc@purelymail.comUser-agent: * Allow: / Disallow: /api/ Disallow: /dashboard Disallow: /set-name Disallow: /auth/ Sitemap: https://crit.md/sitemap.xml
# crit > Local CLI for reviewing AI agent output. Open a browser at localhost, comment inline on markdown plans and code diffs, and feed structured feedback back to the agent without leaving your machine. crit is a single-binary Go CLI. No account required. Works offline. Companion hosted service at crit.md lets you share reviews via URL. ## Get started - [Homepage](https://crit.md): Overview, install options, and quick-start. - [Getting started guide](https://crit.md/getting-started): Step-by-step from install to first review, including agent workflow setup. - [Features](https://crit.md/features): Full list of what crit does. Inline comments, round-to-round diffs, structured JSON output, share links. - [Changelog](https://crit.md/changelog): Release history with what changed and when. Install (pick one): ``` brew install tomasz-tomczyk/tap/crit go install github.com/tomasz-tomczyk/crit/cmd/crit@latest nix run github:tomasz-tomczyk/crit yay -S crit.md-bin ``` `yay -S crit.md-bin` is an AUR package (https://aur.archlinux.org/packages/crit.md-bin), maintained outside this project. Binary downloads on the GitHub releases page. ## Install for your agent For Claude Code, the **plugin marketplace is the recommended path**. It installs the /crit slash command, the crit-cli skill, and a plan-mode hook that intercepts every plan for inline review: ``` claude plugin marketplace add tomasz-tomczyk/crit claude plugin install crit@crit ``` For Codex, **`crit install codex-plugin` is the recommended path**. It installs the $crit skill, crit-cli, and a proposed-plan Stop hook that reviews in-chat plans before the turn ends: ``` cd ~ && crit install codex-plugin ``` For every other agent, `crit install <tool>` writes the integration files (skills, custom rules, or a slash command, depending on the tool). - [All integrations](https://crit.md/integrations): Index of every supported agent with install instructions. - [Claude Code](https://crit.md/integrations/claude-code): Plugin marketplace (recommended, see above) or `crit install claude-code` per project. - [Cursor](https://crit.md/integrations/cursor): `crit install cursor`. Drops the /crit slash command and crit-cli skill into `.cursor/skills/`. - [GitHub Copilot](https://crit.md/integrations/github-copilot): `crit install github-copilot`. Installs as Agent Skills under `.github/skills/`. - [OpenCode](https://crit.md/integrations/opencode): `crit install opencode`. Skill auto-activates on plan and review requests. - [Codex](https://crit.md/integrations/codex): `crit install codex-plugin` globally (recommended, see above) or `crit install codex` per project for skills only. - [Amp](https://crit.md/integrations/ampcode): `crit install ampcode`. Skills under `.agents/skills/` (project) or `~/.config/agents/skills/` (global). - [Windsurf](https://crit.md/integrations/windsurf): `crit install windsurf`. Single rule under `.windsurf/rules/`. - [Cline](https://crit.md/integrations/cline): `crit install cline`. Single rule under `.clinerules/`. - [Aider](https://crit.md/integrations/aider): Append the crit conventions to your `CONVENTIONS.md`. Raw integration files (skills, rules, plugin manifests): https://github.com/tomasz-tomczyk/crit/tree/main/integrations ## Use crit Subcommands an agent will call directly: - `crit` (no args): Auto-detect changed files in the current repo and open them for review. - `crit <file>`: Review a specific file (markdown plan, design doc, code). - `crit comment <path>:<line> <body>`: Add an inline comment programmatically without opening the browser. - `crit comment --reply-to <id> <body>`: Reply to an existing comment. - `crit share <file>`: Publish a review to crit.md and print a share URL. Add `--qr` for a terminal QR code. - `crit unpublish`: Remove a shared review. - `crit pull [pr]`: Fetch GitHub PR review comments into the local review file. - `crit push [pr]`: Post local review comments as a GitHub PR review. Supports `--event approve|request-changes|comment`. - `crit status`: Show review file path and daemon status. ### Story mode For large branch / PR / MR / range diffs, [Story mode](https://crit.md/modes/story) adds a chaptered overview (prologue + thematic chapters + support bucket) before line-by-line review. - Prefer `/crit-story` after `crit install <tool>` - the agent authors the chapters. - Alternative: `crit story` from the terminal when `agent_cmd` is configured (separate LLM spend). - Explicit invoke only - agents must not infer story generation from a normal `/crit` review. - Token cost: LLM-driven exploration on your agent's model. Spend depends on change complexity more than raw file/diff size and does not scale linearly. In our experience, complex PRs (~20–50 files, ~2k–5k lines) cost about $1–$1.40 with Claude Opus 5 - see the Story mode page for a size table. Review files live at `~/.crit/reviews/<key>.json` (keyed by cwd + branch). The JSON schema is documented in the README. Agents read this file to consume human feedback. Humans never edit it directly. Round diffs: when the human reopens crit after the agent makes changes, crit shows only what changed since the last review round, not the full diff again. ## Source - [CLI source (Go)](https://github.com/tomasz-tomczyk/crit): Single binary, local review server, round-diff engine, integration installer. - [CLI README](https://github.com/tomasz-tomczyk/crit/blob/main/README.md): Full usage reference. Subcommands, flags, review file schema, integration setup. - [Hosted service source (Elixir/Phoenix)](https://github.com/tomasz-tomczyk/crit-web): The share-via-URL backend. Same review UI as the local CLI. ## Optional - [Self-hosting](https://crit.md/self-hosting): Run your own crit-web instance. Useful for teams that can't send code to a third-party service. Point the CLI at your instance with `--share-url` or the `share_url` config key.
| – |
| 7 Nov 2026 |
| – |
| 7 Nov 2026 |