Codex MCP Setup

Use this setup when Codex should discover TextMine docs and API schemas through ReadMe MCP, or call the TextMine MCP product tools with a TextMine API key.

Which endpoint to use

Use https://mcp.textmine.com/mcp for public TextMine MCP setup. Use environment-specific MCP hosts only for private preview, staging validation, or dedicated deployments when TextMine explicitly issues that host.

Prerequisites

• A TextMine API key for the user, integration, or third-party agent service account.
• Codex with remote HTTP MCP support.

If an autonomous third-party agent does not yet have a key, use Agent self-service signup. Signup is REST-first and returns an encrypted tm_... service-account key for the trusted runtime to store.

Recommended setup

Store the API key in an environment variable:

export TEXTMINE_API_KEY="tm_your_api_key"

Add the ReadMe MCP server for docs and generated API-reference tools:

codex mcp add textmine-docs \
  --url https://developers.textmine.com/mcp

Add the TextMine MCP server when Codex should use product tools directly:

codex mcp add textmine \
  --url https://mcp.textmine.com/mcp \
  --bearer-token-env-var TEXTMINE_API_KEY

If your Codex MCP configuration uses generated API-reference calls through ReadMe MCP, attach the same bearer token to that server as well. For docs discovery only, the ReadMe MCP server does not need a TextMine product key. If a client cannot set Authorization, use X-TextMine-Api-Key with the same tm_... key.

Check the servers are configured:

codex mcp list

First prompt to try

Use TextMine docs to read the For Agents page, list the relevant Public API V3 operation IDs, then explain the safe sequence for importing a Google Drive file into Vault.

Operating guidance

Codex should use listVaults and other discovery calls before mutations, and should ask before destructive changes, new workflows, scheduled agents, provider disconnects, or credential creation.

Keep tm_... keys in environment variables or secure credential storage. Do not paste real keys into prompts, repo files, task transcripts, or shared docs.