Skip to content

OpenAI Codex

Guide for connecting codex with PlanExe via MCP.

Prerequisites

  • Access to Codex.
  • PlanExe MCP server reachable by Codex.

Quick setup

  1. Start Codex.
  2. Ask for MCP tools.
  3. Call example_prompts to get examples.
  4. Call plan_create to start a plan.

Sample prompt

Get example prompts for creating a plan.

Success criteria

  • You can retrieve prompt examples.
  • You can create a plan task.
  • You can download the report.

Interaction

In a terminal, start codex like this:

codex

Inside codex; these are my interactions:

  1. tell me about the mcp tools you have access to
  2. for planexe, get the prompt examples
  3. I want you to formulate a prompt about constructing a new variant of english where the worst inconsistencies have been fixed such as 11th vs 1st, 21st, 31st, potentially eliminated such suffixes. And the pronounciation inconsistencies have been cleaned up. I want you to adhere to the planexe example prompts.
  4. it's not just the ordinals. try again
  5. go ahead create this plan
  6. status
  7. status
  8. status
  9. download both
  10. summarize the html file

The created plan is here: Clean English

Prerequisites

A working installation of PlanExe.

  • The recommended way is to install PlanExe by following the Getting Started instructions. Make sure that docker compose up is running, in order to connect to PlanExe.
  • Alternatively: Run PlanExe on another server and port.
  • Alternatively: If you are a developer run PlanExe inside a python virtual environment.

Double check that PlanExe can take a prompt and create a plan. Since it doesn't make sense to start configuring Cursor if the PlanExe installation is incomplete.

Configuring Codex

OpenAI's MCP documentation

Point Codex at the PlanExe MCP server over HTTP (cloud or local Docker).

Cloud server:

codex mcp add planexe --url https://mcp.planexe.org/mcp --header "X-API-Key: pex_YOUR_API_KEY"

Local Docker:

codex mcp add planexe --url http://localhost:8001/mcp

Adjust the URL if PlanExe is running on another port.

Verify that it's working.

codex mcp list