Co-Review
An IDE-style code review tool that takes an entire code repository as its review object. It supports agent-assisted repository audits, PR reviews and design planning, and emphasizes inline comments and threaded discussion interaction.
Timeline (last 90 days)
Co-Review is an IDE tool that lets users open an entire code repository as they would in an IDE, treat every line of code as reviewable, and ask the agent questions or leave notes anywhere.
Co-Review offers four reusable review scenarios: a whole-codebase audit, reviewing the agent's own changes, reviewing someone else's Pull Request, and designing before writing code.
In the audit scenario, Co-Review's audit agent first reads the entire codebase, divides it into regions and flags findings, while tracking the coverage of what the user has already read.
When reviewing the agent's changes, Co-Review reuses the Pull Request inline-comment interaction model: the user comments on a line, the agent replies in the thread and requests changes, and then each comment is fixed as a new commit with a reply in each thread.
Co-Review supports reviewing someone else's Pull Request: the user provides the PR number, the agent reads and checks out that PR's code and completes a first round of review, and the user can add inline comments and publish them to GitHub.
Co-Review supports designing before writing code: after the user describes the task, the agent investigates and writes a plan including diagrams and risk analysis, the user can comment on steps or diagram nodes, and after approval the agent builds only what was agreed.
Co-Review is provided as a plugin, keeping Claude in the session so that the user's comments arrive in the session in the background, allowing the user to keep responding as reviewer while Claude fixes the problems.
Besides the Claude Code plugin, Co-Review also provides an API and native commands, so that Codex, OpenCode, Cursor, Gemini, Goose and any MCP agent can connect and get the same four review capabilities.
Getting started with Co-Review from Claude Code requires adding a marketplace and installing a plugin, four commands in total.
The core value of Co-Review lies in extending the granularity of review from the diff to the entire repository, letting the agent take on the first round of screening while people focus on high-risk items and key decisions.
All Facts (10)
Co-Review is an IDE tool that lets users open an entire code repository as they would in an IDE, treat every line of code as reviewable, and ask the agent questions or leave notes anywhere.
50%UnverifiedCo-Review offers four reusable review scenarios: a whole-codebase audit, reviewing the agent's own changes, reviewing someone else's Pull Request, and designing before writing code.
50%UnverifiedIn the audit scenario, Co-Review's audit agent first reads the entire codebase, divides it into regions and flags findings, while tracking the coverage of what the user has already read.
50%UnverifiedWhen reviewing the agent's changes, Co-Review reuses the Pull Request inline-comment interaction model: the user comments on a line, the agent replies in the thread and requests changes, and then each comment is fixed as a new commit with a reply in each thread.
50%UnverifiedCo-Review supports reviewing someone else's Pull Request: the user provides the PR number, the agent reads and checks out that PR's code and completes a first round of review, and the user can add inline comments and publish them to GitHub.
50%UnverifiedCo-Review supports designing before writing code: after the user describes the task, the agent investigates and writes a plan including diagrams and risk analysis, the user can comment on steps or diagram nodes, and after approval the agent builds only what was agreed.
50%UnverifiedCo-Review is provided as a plugin, keeping Claude in the session so that the user's comments arrive in the session in the background, allowing the user to keep responding as reviewer while Claude fixes the problems.
50%UnverifiedBesides the Claude Code plugin, Co-Review also provides an API and native commands, so that Codex, OpenCode, Cursor, Gemini, Goose and any MCP agent can connect and get the same four review capabilities.
50%UnverifiedGetting started with Co-Review from Claude Code requires adding a marketplace and installing a plugin, four commands in total.
50%UnverifiedThe core value of Co-Review lies in extending the granularity of review from the diff to the entire repository, letting the agent take on the first round of screening while people focus on high-risk items and key decisions.
50%