Skip to main content
Monocle is a terminal UI that runs alongside your AI coding agent. You review diffs in real time as the agent writes code, leave line-level comments — issues, suggestions, notes — and submit a structured review in one batch. The agent receives your feedback and starts fixing things immediately, just like a PR review but live.

Installation

Install Monocle via Homebrew, pre-built binaries, or from source

Quickstart

Register Monocle with your agent and start your first review session

Review Loop

Understand how the live review cycle works end-to-end

CLI Reference

Complete reference for all Monocle commands and flags

​
How it works

Monocle connects to your AI coding agent using Unix sockets. You run Monocle in a separate terminal while your agent works. As the agent writes code, you see diffs appear in real time, leave line-level comments, and submit your review when you’re ready. The agent retrieves your feedback and starts addressing it — you review the updated diffs, iterate, and ship.
1

Install Monocle

Install via Homebrew or download a pre-built binary for your platform.
brew install josephschmitt/tap/monocle
2

Register with your agent

Run monocle register to install skills into your agent’s config directory. This works with Claude Code, OpenCode, Codex CLI, and Gemini CLI.
monocle register          # interactive picker
monocle register claude   # or: opencode, codex, gemini, all
3

Start reviewing

Launch your agent in one terminal, then start Monocle in another. Diffs appear automatically as the agent writes code.
monocle
4

Submit feedback

Navigate diffs with j/k, add comments with c, and press S to submit your review. The agent receives your feedback and starts fixing things.

​
Key features

Live diff viewer

Unified and split (side-by-side) views with syntax highlighting and intra-line diffs. Switch styles with t.

Structured comments

Tag feedback as issues, suggestions, notes, or praise. Add line-level or file-level comments with visual range selection.

Plan review

Review your agent’s plans and architecture decisions before it writes code. Block the agent until you approve with /review-plan-wait.

Push notifications

With Claude Code channels, feedback is pushed directly into the agent’s context the moment you submit — no polling needed.

Suggested edits

Propose exact code changes with GitHub-style suggestion blocks that the agent can apply directly.

Session persistence

Reviews survive restarts via SQLite. View past submission history with :history.

​
Supported agents

Monocle works with any coding agent that supports agent skills, including:
  • Claude Code — Full support including push notifications via MCP channels
  • OpenCode — Pull-based feedback via skills
  • Codex CLI — Pull-based feedback via skills
  • Gemini CLI — Pull-based feedback via skills

Build docs developers (and LLMs) love

Get started for freeTalk to us