What is Agentic Workbench Mcp

Agentic Workbench MCP (AgenticBench) is a server that leverages OpenAI’s Codex to facilitate structured development workflows, allowing developers to control AI agents for coding tasks in a recursive manner.

Use cases

Use cases for AgenticBench include developing web application components, creating service classes in C#, and any scenario where consistent and accurate code generation is required through AI assistance.

How to use

To use AgenticBench, developers should follow three steps: 1) Utilize Codex as the AI code writer, 2) Prepare predefined ‘workbenches’ for consistent project structures, and 3) Connect these workbenches recursively to enhance component creation.

Key features

Key features of AgenticBench include integration with OpenAI’s Responses API, support for local and hosted models, tailored environments for AI agents, and a structured approach to coding that ensures repeatable accuracy.

Where to use

AgenticBench can be utilized in software development, particularly in environments where AI-assisted coding is beneficial, such as web development, service-oriented architecture, and component-based design.

Agentic Developer MCP

This project wraps OpenAI’s Codex CLI as an MCP (Model Context Protocol) server, making it accessible through the TeaBranch/open-responses-server middleware.
This engine may be replaced with OpenCode or Amazon Strands

Requirements

• Node 22 (nvm install 22.15.1 | nvm use 22.15.1) required for Codex

Overview

The setup consists of three main components:

1. Codex CLI: OpenAI’s command-line interface for interacting with Codex.
2. MCP Wrapper Server: A Node.js Express server that forwards MCP requests to Codex CLI and formats responses as MCP.
3. open-responses-server: A middleware service that provides Responses API compatibility and MCP support.

Installation

Using Docker (Recommended)

# Clone this repository
git clone https://github.com/yourusername/codex-mcp-wrapper.git
cd codex-mcp-wrapper

# Start the services
./start.sh

This will start:

• Codex MCP wrapper on port 8080
• open-responses-server on port 3000

Manual Installation

# Install dependencies
npm install

# Install Codex CLI globally
npm install -g @openai/codex

# Start the MCP server
node mcp-server.js
# Install the package in development mode
pip install -e .

Usage

You can run the MCP server using either stdio or SSE transport:

# Using stdio (default)
python -m mcp_server

# Using SSE on a specific port
python -m mcp_server --transport sse --port 8000

Tool Documentation

run_codex

Clones a repository, checks out a specific branch (optional), navigates to a specific folder (optional), and runs Codex with the given request.

Parameters

• repository (required): Git repository URL
• branch (optional): Git branch to checkout
• folder (optional): Folder within the repository to focus on
• request (required): Codex request/prompt to run

Example

{
  "repository": "https://github.com/username/repo.git",
  "branch": "main",
  "folder": "src",
  "request": "Analyze this code and suggest improvements"
}

clone_and_write_prompt

Clones a repository, reads the system prompt from .agent/system.md, parses modelId from .agent/agent.json, writes the request to a .prompt file, and invokes the Codex CLI with the extracted model.

Parameters

• repository (required): Git repository URL
• request (required): Prompt text to run through Codex
• folder (optional, default /): Subfolder within the repository to operate in

Example

{
  "repository": "https://github.com/username/repo.git",
  "folder": "src",
  "request": "Analyze this code and suggest improvements"
}

MCPS Configuration

Place a mcps.json file under the .agent/ directory to register available MCP tools. Codex will load this configuration automatically.

Example .agent/mcps.json:

{
  "mcpServers": {
    "agentic-developer-mcp": {
      "url": "..."
    }
  }
}

Development

This project uses the MCP Python SDK to implement an MCP server. The primary implementation is in mcp_server/server.py.

License

MIT