Ask your AI about your CodeMouse reviews

Connect CodeMouse to Claude or ChatGPT over the Model Context Protocol (MCP) and ask, in plain language, about your accounts, repositories, and pull-request reviews — what passed, what failed, how long it took, and what it cost.

Ask it things like

• Which of my PRs failed review this week, and why?
• How much have my CodeMouse reviews cost this month?
• Show me the last review on acme/api and what it flagged.
• Which repositories does CodeMouse have access to?
• What was the longest-running review, and how long did it take?

Server URL

https://codemouse.ai/api/mcp

Connect it

ChatGPT (web / desktop)

CodeMouse is in the ChatGPT plugin directory — open the listing, add the plugin, then sign in with GitHub and approve.

Claude (web / desktop / mobile)

CodeMouse is in the Claude directory — open the listing, click Connect, then sign in with GitHub and approve.

Any other MCP client: add a custom connector / MCP server with https://codemouse.ai/api/mcp.

Claude Code

claude mcp add --transport http codemouse https://codemouse.ai/api/mcp

Then run /mcp inside Claude Code to finish the GitHub sign-in.

Codex CLI

Add to ~/.codex/config.toml:

[mcp_servers.codemouse]
command = "npx"
args = ["-y", "mcp-remote", "https://codemouse.ai/api/mcp"]

Cursor & other stdio tools

Use the mcp-remote bridge, which handles the OAuth browser flow: npx -y mcp-remote https://codemouse.ai/api/mcp

What it can do

• list_accounts — the CodeMouse accounts you belong to, with your role and subscription status.
• list_repositories — the GitHub orgs and repositories CodeMouse can access, per account.
• list_reviews — recent PR reviews for an account: status, PR, repo, duration, cost.
• get_review — full detail of one review: status, timing, cost, tokens, per-provider runs, errors.

Authentication uses OAuth 2.1 (PKCE). CodeMouse never sees your AI provider keys, and the connector can only read data for accounts where you are a member or owner.