Arlet
GitHub

Connect GitHub to Codex

Search, read and write GitHub from Codex. Arlet links the two through one MCP connection and takes care of keys and authentication.

Enable GitHub in ArletCodex connection guide (docs)

Arlet makes GitHub's official remote MCP server available from the AI client you already use. Signing in with your GitHub account completes the connection.

The AI can read repositories and files, create, update and review issues and pull requests, inspect GitHub Actions and search code. The available tools are those provided by GitHub's official server, and what can be read or written is exactly what your GitHub account can. If your organization restricts OAuth app access, ask an organization owner to approve Arlet.

How to connect GitHub to Codex

  1. 1

    In the Arlet dashboard, open Integrations → Add an integration and choose GitHub. Sign in with Link GitHub, name the connection and click Create.

  2. 2

    In your terminal, run codex mcp add arlet --url https://mcp.arlet.dev/mcp/ to add Arlet as an MCP server.

  3. 3

    Codex detects OAuth support and opens your browser. Sign in with Google and click Allow on the consent screen.

  4. 4

    Once connected, both the CLI and the IDE extension can call Arlet tools (the configuration is shared).

If the browser does not open, run codex mcp login arlet to redo the authentication.

What you can do

GitHub tools that Codex can call.

  • get_me

    Get the connected user's profile.

  • search_code

    Search code across repositories.

  • get_file_contents

    Read a file or directory in a repository.

  • list_commits

    List commits.

  • create_or_update_file

    Create or update a file and commit it.

  • push_files

    Commit several files at once.

  • list_issues

    List issues.

  • issue_read

    Read an issue's details and comments.

  • issue_write

    Create or update an issue.

  • add_issue_comment

    Comment on an issue or pull request.

  • search_issues

    Search issues.

  • create_pull_request

    Open a pull request.

  • pull_request_read

    Read a pull request's details, diff and reviews.

  • search_pull_requests

    Search pull requests.

  • merge_pull_request

    Merge a pull request.

  • actions_list

    List GitHub Actions workflows and runs.

  • get_job_logs

    Read Actions job logs.

  • list_notifications

    List notifications.

The tool list follows what GitHub's official MCP server provides and may change.

Example uses

  • Review a pull request

    Read the diff and comments, point out problems and post them as review comments.

  • Debug a failed workflow

    Read the failing job's log and summarize the cause and a fix.

  • File issues

    Turn meeting notes or bug reports into issues with labels and assignees.

  • Explore the code

    Search across repositories to find where something is implemented and explain it.

Prompts you can use as-is

  • Review the diff of PR #1836 in NexaScience/arlet and post your concerns as comments
  • Read the job log of the last failed Actions run on main and explain the cause
  • Rewrite this bug report as steps, expected and actual results, and open an issue with the bug label
  • Find where payment retries are implemented in the repository and explain it

Use GitHub with other AI clients