easy-review documentation
Review diffs at the speed of thought
easy-review (the er binary) is a git diff review tool built for
developers who work with AI coding assistants. AI writes code faster than you can read it —
er makes review fast, navigable, and live-updating, whether you prefer a keyboard-driven
terminal UI or a full desktop application.
New here? Start with Installation, then walk through your first review in the Quick Start. If you just want the key map, jump to the Keybinding Reference.
Two front ends, one engine
er ships as two applications that share the same review engine, the same on-disk review data,
and the same core concepts. Pick the one that fits how you work — or use both on the same repository and
branch, since they read and write the same review files.
The er TUI
A single, self-contained binary. Keyboard-driven, instant startup, runs anywhere a terminal does — perfect next to Claude Code in a split pane.
The desktop app
A Tauri application with split diffs, an embedded terminal and browser, multi-model review “arena”, and a settings GUI.
What you can do with it
- Read diffs in six modes — branch, unstaged, staged, commit history, merge conflicts, and hidden/ignored files, plus a dedicated PR diff view and an AI-generated guided tour.
- Follow changes live — watch mode refreshes the diff as files are saved, staged, and committed.
- Track your progress — mark files reviewed, filter to what is left, jump to the next unreviewed file.
- Leave three kinds of notes — private questions for yourself or your AI, private notes for agent hand-off, and GitHub comments that sync to a pull request.
- Layer in AI analysis — per-file risk levels, inline findings, a review checklist, and specialized expert agents.
- Work with pull requests — open a PR by number or URL, see CI and reviewer status, and sync comments two ways.
How this documentation is organized
Most features work the same in both the terminal and desktop apps. Those shared ideas live under Core Concepts. Where the two differ, the dedicated Terminal UI and Desktop App sections cover what is unique to each. Throughout the docs, look for these badges:
both works in the terminal and desktop apps terminal terminal UI only desktop desktop app only
| If you want to… | Read |
|---|---|
| Install either app | Installation |
| Run your first review end to end | Quick Start |
| Wire Claude / Cursor / Codex to Easy Review | MCP Server Setup |
Open er inside Herdr worktrees | Herdr Integration |
| Understand the mental model | How er Works |
| Learn the diff modes | Diff Modes |
| Comment and ask questions | Comments & Questions |
| Set up the AI review loop | AI Review & AI Hub Actions |
| Connect a GitHub pull request | GitHub & Pull Requests |
| Customize behavior | Configuration |
| Memorize the keyboard | Keybinding Reference |
A note on your data
er is a viewer and orchestrator. It does not run AI models itself, and it does
not transmit your code anywhere. There is no telemetry, no analytics, and no backend — it shells out to your
local git (and optionally the gh CLI) and reads review artifacts that AI tools
you run write to disk. You bring your own AI tools and credentials. See
AI Review for how that loop works.