Review the live worktree
GitWarren reads the active worktree directly, including staged, unstaged, and untracked files, so you can inspect the full set of changes before a commit exists.
GitWarren is a desktop app for local Git code review, combining staged, unstaged, and untracked changes before a commit or pull request, with an MCP server for local AI agents.
GitWarren is a desktop app for reviewing Git worktrees locally before anything is committed or pushed. It is aimed at the gap between an agent finishing a task and the point where its changes become a pull request, so you can inspect the diff on your own machine first.
The app reads staged, unstaged, and untracked files directly from the worktree and combines them into a single review view. It also ships an MCP server over stdio so local AI agents can take part in the review by reading discussions, replying in threads, and leaving or resolving comments.
GitWarren reads the active worktree directly, including staged, unstaged, and untracked files, so you can inspect the full set of changes before a commit exists.
The app folds the worktree into one diff and lets you read and comment on files that have not yet been committed or pushed.
An MCP server over stdio exposes the same review tools to local agents such as Claude Code and Codex, letting them participate in the discussion.
Agents can open a review, read the thread history, reply in a thread, leave line comments, and resolve comments using the same review surface.
Each machine-written comment is marked as machine-written, and the tool name comes from the MCP handshake rather than the model itself.
Reviews and comments are stored locally in one SQLite file in the application-data directory, while branch names, commits, and diffs are read from git when shown.
Use GitWarren when an agent finishes a task and leaves a dirty worktree. You can inspect the changes immediately, before committing or turning them into a pull request.
Use it for local code review on your own machine when you want to see staged, unstaged, and untracked files together in one diff rather than reviewing only committed work.
Use the MCP server when you want an AI agent to respond to review feedback directly inside the discussion, rather than in a separate chat or external tool.
Use GitWarren when you want review history, comments, and repository context to remain local instead of being stored in a hosted service or shared account.
GitWarren is a desktop app for local Git code review. It reads the worktree directly, so you can review staged, unstaged, and untracked files before anything is committed or opened as a pull request.
The desktop app is available for macOS, with downloads labeled for Apple silicon and Intel. The site also mentions Windows and Linux support, and Linux packages are available as deb and rpm.
GitWarren is designed to work locally with no account, no sign-in, and no server. Reviews, comments, and repository data stay on your machine in a single SQLite file.
Yes. GitWarren includes an MCP server over stdio so local AI agents can open reviews, read discussions, reply in threads, leave line comments, and resolve comments through an MCP client.
The app is free software released under GPL-3.0, and the site also says it is free and open source. A pricing page exists on the site but currently returns a not-found page, so no official pricing details are provided.
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.
Vi3ecode is a maintained development environment for AI coding agents, keeping project context, terminal, Git, memory, flows and review connected.
Lucid Train is a local-first AI coding harness that turns repo code into architecture diagrams for coding agents. Desktop app and Rust CLI, local models or API keys.
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.