Review Every Change Your AI Agents Made
A full-screen diff viewer built into AgentsRoom. See what changed, ask questions about the code, and commit when you are satisfied. File by file, with syntax highlighting and word-level diffs.
What Review Mode Includes
Every tool you need to inspect AI-generated changes before committing them to your repository.
Side-by-Side Diff Viewer
View old and new code side by side with syntax highlighting via highlight.js. Word-level diffs use an LCS algorithm to show exactly which tokens changed within each line.
Color-Coded File List
Left sidebar lists every modified file. Green for added, red for deleted, orange for modified. Click any file to jump to its diff. Track your progress with the review bar.
Keyboard Navigation
Navigate between files with arrow keys (left/right) or j/k. Mark files as reviewed with the eye icon. Audio feedback plays a subtle tick when navigating and a chime at the last file.
Comment on Any Line
Click Comment on this line anywhere in the diff and write what should change. Your comments pile up as you review, then Send comments queues them for the agent as one message, so it picks them up when its current turn ends.
Progress Tracking
A progress bar shows how many files you have reviewed out of the total. Mark files reviewed as you go. The completion screen appears only when every file has been checked.
Commit When Ready
Once all files are reviewed, a completion screen shows a commit message textarea. Generate a commit message with AI or write your own. Click commit, get success or error feedback, and the modal auto-closes.
How It Works
Four steps from agent output to committed code.
Agent finishes work
One or more Claude agents complete their tasks and modify files in your project. AgentsRoom detects the changed files automatically.
Open Review Mode
Click the review button to open a full-screen modal. The file list loads in the sidebar with color-coded status indicators for each changed file.
Review file by file
Navigate through files using keyboard shortcuts or the sidebar. Inspect diffs, leave a comment on any line that needs one, and mark each file as reviewed.
Commit all changes
When every file is marked reviewed, the completion screen appears. Write or generate a commit message, then commit. The modal closes on success.

What diff format does Review Mode use?
Can I ask questions about the code during review?
How does the commit flow work?
What keyboard shortcuts are available?
Do I have to review every file before committing?
Further reading
Should You Still Review Your AI Agent's Code?
Your agents write better code than half the pull requests you used to merge. So do you still read every line? The honest case for both sides, the 10 signals that tell you an agent messed up, and how much review each change actually deserves.
Claude Watermarks Its Output Now. Your Code Is Barely Affected.
Anthropic now watermarks Claude's output. What it really marks, why generated code mostly escapes it, who can actually detect it, and why your SEO does not change.
Review AI Code Changes With Confidence
Download AgentsRoom and inspect every file your agents modify before it reaches your repository. Free for up to 2 projects.
Companion app: monitor your agents on the go
Bring your own: Claude, Codex, Antigravity CLI, or other AI provider.
Push bugs and requests straight to your public backlog.