Reado — MCP server for Claude Code: hands your review notes to the agent
Lets the Claude agent pick up the tasks you left as comments in Reado and work through them while you watch the diff.
Open on GitHubWhat Reado does
Reado is an editor aimed at reading code rather than typing it. Its MCP server exposes the review side of your project to an agent: open comment tasks, the comments themselves, and the line ranges they are pinned to. The agent walks those tasks, pulls in docs and specs, and takes them one at a time. Comments sit as plain Markdown inside .reado/ next to your code, so they travel with the repository.
Each comment has a type — bug, refactor, performance, question, note. It is anchored to a range of lines and follows them when the code shifts; when the lines go away, the comment becomes an orphan instead of quietly landing somewhere wrong. So the loop is: you mark up a diff, the agent resolves each item through the reado CLI and MCP server, then you check what changed.
You need the Reado app itself running with a project open. Without it the server has nothing to serve. Setting it up for an MCP client from the supported list comes down to that client's own configuration.
What it helps with
- hand the agent a list of diff remarks with their exact line anchors
- split a review into typed tasks such as bug or refactor
- spot which files changed since you last read them
- keep a comment attached after the surrounding code moves
- give the agent the project docs and specs in one searchable place
Example prompts for the agent
- List the open tasks in
.reado/and start with the riskiest one - Resolve the bug task on
parser.tsand show me the diff - What changed in this file since I last read it?
How to install Reado
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Reado
How do I install the Reado MCP server in Claude Code?
There is no separate install command: the server ships with the Reado application. Download and open the editor, load a project, then set up the MCP connection to your agent inside the app.
What does Reado do?
It gives an agent your review comments and tasks as context and tools, so it can read them and close them out. You still decide what gets accepted.
Does it need a real repository to run?
Yes. The server works against a project open in Reado, and comments live in .reado/ anchored to code lines. With no code there is nothing to resolve.
