Codex MCP Server

Run parallel, resumable, human-operable Antigravity CLI sessions from any MCP agent harness

Local serverstdioPython

What is the Codex MCP server?

Codex MCP server exists for a simple reason — assistants are far more useful when they can act on Codex directly instead of describing what you should do. Run parallel, resumable, human-operable Antigravity CLI sessions from any MCP agent harness.

Setting it up

The server ships on PyPI as codex-agy-bridge, so your MCP client can launch it on demand — there is no separate build step. Add the server block to your client's configuration, restart it, and the tools register themselves.

What the assistant can call

Once Codex is connected, these are the calls the assistant has available:

  • agy_run_start — Start, continue, or open an interactive foreground run
  • agy_run_wait — Short-poll until selected runs emit sparse wake events
  • agy_run_observe — Read full, status, transcript, or raw terminal views
  • agy_run_input — Send input with optional event/transcript preconditions
  • agy_run_cancel — Cancel one active run
  • agy_run_result — Read final result metadata or bounded result chunks
  • agy_goal — Create goals, start targets, and read aggregate status
  • agy_admin — Read diagnostics, models, plugins, validation, and changelog
  • PyPI — The Quick Install command stores an stdio MCP server definition. When the agent harness starts the server, uvx resolves codex-agy-bridge@latest from
  • GitHub — The GitHub tool exposed by this server

Before you rely on it

  • It runs with your machine's permissions. That is convenient and also the reason to think about what you point it at before you approve a tool call.
  • With 10 tools registered it takes up a noticeable share of the context window; turn it off in projects that never touch Codex.
  • MCP clients confirm each tool call by default. Leave that on until you have watched what the codex mcp server does with a few real requests.

Choosing this one

This sits in the developer tooling group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. Codex's toolset — agy_run_start, agy_run_wait, agy_run_observe and 7 more — is a fair guide to whether it matches your workflow. It is maintained by Varad; worth a glance at recent repository activity before you build anything load-bearing on it.

This entry was verified against Codex's own documentation before publication; SyncDev keeps the directory reviewed rather than auto-generated.

Available tools

ToolWhat it does
agy_run_startStart, continue, or open an interactive foreground run
agy_run_waitShort-poll until selected runs emit sparse wake events
agy_run_observeRead full, status, transcript, or raw terminal views
agy_run_inputSend input with optional event/transcript preconditions
agy_run_cancelCancel one active run
agy_run_resultRead final result metadata or bounded result chunks
agy_goalCreate goals, start targets, and read aggregate status
agy_adminRead diagnostics, models, plugins, validation, and changelog
PyPIThe Quick Install command stores an stdio MCP server definition. When the agent harness starts the server, uvx resolves codex-agy-bridge@latest from PyPI, installs it into an isolated cached environment, and runs the cod
GitHubThe GitHub tool exposed by this server.

How to install the Codex MCP server

{
  "mcpServers": {
    "codex-agy-bridge": {
      "command": "uvx",
      "args": ["codex-agy-bridge"]
    }
  }
}

Add to claude_desktop_config.json, then restart Claude Desktop.

Example prompts to try

  • Use Codex to agy run start.
  • Use Codex to agy run wait.
  • Use Codex to agy run observe.

Frequently asked questions

It connects Codex to MCP-compatible AI assistants such as Claude and Cursor, exposing 10 tools (agy_run_start, agy_run_wait, agy_run_observe, and more) that the assistant can call on your behalf. Instead of copying data back and forth by hand, the assistant works with Codex directly.