crit

Crit lets you view and comment on the output of AI coding agents directly in a browser. It works locally, so your data stays on your machine, and it can handle plans, code diffs, running apps, and static HTML artifacts. The interface is designed to replace scrolling through terminal diffs with a visual review surface.

The tool expects an agent to write its output to a file and then invoke a command that launches Crit. Once opened, Crit detects the type of artifact, markdown plan, multi‑file diff, live dev server, or static HTML, and presents an appropriate view. For plans and documentation written in markdown, Crit renders the text with a per‑section comment box, letting you target feedback without editing the whole file. When an agent produces a code diff, Crit auto‑detects the changed files, applies syntax highlighting, and lets you add line‑level comments similar to a pull‑request review, but without needing a remote repository. If the agent runs a frontend on localhost, Crit proxies the page into an iframe where you can click UI elements and pin comments directly on the DOM. Static HTML files are also shown in an iframe, allowing interaction without a live server.

Crit is built to be agent‑agnostic. Any AI coding agent that can write files and execute a command can be integrated, because the communication happens through a structured review file rather than an API or SDK. The site lists a number of ready‑made integrations for popular agents such as Claude Code, GitHub Copilot, Gemini CLI, and others; each can be added with a simple install command. Because the software is open source and can be installed via Homebrew (`brew install crit`) or from source, you can also self‑host it behind your own firewall.

Key features

• Local‑first operation; no cloud service required
• Browser‑based rendering of markdown plans with per‑section comments
• Syntax‑highlighted, multi‑file code diffs with line‑level commenting
• Interactive review of running apps by proxying localhost pages
• Iframe preview of static HTML artifacts
• Plug‑in style integrations for many existing AI coding agents