Codex MCP Go

claude mcp add codex -s user --transport stdio -- npx -y @zenfun510/codex-mcp-go

Add the following to the MCP settings in Roo Code:

{
  "mcpServers": {
    "codex": {
      "command": "npx",
      "args": ["-y", "@zenfun510/codex-mcp-go"],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

Refer to the following paths for the configuration file:

• VSCode: ~/.config/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json
• Cursor: ~/.config/Cursor/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json

Add the following to ~/.kilocode/mcp.json:

{
  "mcpServers": {
    "codex": {
      "command": "npx",
      "args": ["-y", "@zenfun510/codex-mcp-go"],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

1. Open Cursor settings -> Features -> MCP
2. Click "Add New MCP Server"
3. Fill in the configuration:
   • Name: codex
   • Type: stdio
   • Command: npx
   • Args: -y @zenfun510/codex-mcp-go

claude mcp add codex -s user --transport stdio -- /path/to/codex-mcp-go

{
  "mcpServers": {
    "codex": {
      "command": "/path/to/codex-mcp-go",
      "args": [],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

1. Open Cursor settings -> Features -> MCP
2. Click "Add New MCP Server"
3. Fill in the configuration:
   • Name: codex
   • Type: stdio
   • Command: /path/to/codex-mcp-go
   • Args: (Leave empty)

## Core Instruction for CodeX MCP

At any time, you must think about how the current process can collaborate with codex and how to call the MCP tools provided by Codex as a guarantee for your objective and comprehensive analysis.
Among them, you **must execute** the following steps:
**1** After you form a preliminary analysis of the user's needs, inform codex of the user's needs and initial ideas, and ask it to improve the needs analysis and implementation plan.
**2** Before implementing a specific coding task, **you must ask codex for a code implementation prototype (require codex to only provide a unified diff patch and严禁 make any real modifications to the code)**. After obtaining the code prototype, **you can only use it as a logical reference and rewrite the code modifications again**, and then implement the specific programming modification task after forming enterprise-level, highly readable, and highly maintainable code.
**3** Whenever you complete a real coding action, **you must immediately use codex to review the code changes and the completion of the corresponding requirements**.
**4** Codex can only provide references. **You must have your own thinking and even question codex's answers**. Trusting books too blindly is worse than having no books. The ultimate mission of you and codex is to reach a unified, comprehensive, and accurate opinion, so you must constantly argue to find the only way to truth.

## Codex Tool Invocation Specification

1. Tool Overview
   Codex MCP provides a tool called `codex` for performing AI-assisted coding tasks. This tool is **invoked through the MCP protocol** and does not require the use of the command line.

2. Tool Parameters
   **Required** parameters:
   - PROMPT (string): The task instruction sent to codex
   - cd (Path): The root path of the working directory where codex executes tasks

   Optional parameters:
   - sandbox (string): Sandbox strategy, optional values:
     - "read-only" (default): Read-only mode, the safest
     - "workspace-write": Allows writing in the workspace
     - "danger-full-access": Full access permission
   - SESSION_ID (UUID | null): Used to continue the previous session for multi-turn interaction with codex, default is None (start a new session)
   - skip_git_repo_check (boolean): Whether to allow running in a non-Git repository, default is False
   - return_all_messages (boolean): Whether to return all messages (including inferences, tool calls, etc.), default is False
   - image (List[Path] | null): Attach one or more image files to the initial prompt, default is None
   - model (string | null): Specify the model to use, default is None (use the user's default configuration)
   - yolo (boolean | null): Run all commands without approval (skip the sandbox), default is False
   - profile (string | null): The name of the configuration file loaded from `~/.codex/config.toml`, default is None (use the user's default configuration)

3. Invocation Specification
   **Must be followed**:
   - Every time you call the codex tool, you must save the returned SESSION_ID for subsequent conversations.
   - The cd parameter must point to an existing directory; otherwise, the tool will fail silently.
   - It is strictly forbidden for codex to make actual modifications to the code. Use sandbox="read-only" to avoid accidents and ask codex to only provide a unified diff patch.

   Recommended usage:
   - If you need to track the inference process and tool calls of codex in detail, set return_all_messages=True.
   - For tasks such as precise positioning, debugging, and rapid writing of code prototypes, give priority to using the codex tool.

# Codex MCP Tool Rules

You have access to the `codex` tool, which wraps the OpenAI Codex CLI. Use it for complex code generation, debugging, and analysis.

## Workflow
1.  **Consultation**: Before writing complex code, ask Codex for a plan or analysis.
2.  **Prototyping**: Ask Codex for a `unified diff patch` to solve the problem.
    *   **IMPORTANT**: Always use `sandbox="read-only"` when asking for code.
    *   **IMPORTANT**: Do NOT let Codex apply changes directly.
3.  **Implementation**: Read the Codex-generated diff, understand it, and then apply the changes yourself using your own file editing tools.
4.  **Review**: After applying changes, you can ask Codex to review the code.

## Tool Usage
-   **Session**: Always capture and reuse `SESSION_ID` for multi-turn tasks.
-   **Path**: Ensure `cd` is set to the current workspace root.
-   **Safety**: Default to `sandbox="read-only"`. Only use `workspace-write` if explicitly instructed by the user and you are confident in the operation.