Polyagent

MCP server that brings Claude Code agents to Codex, Gemini CLI and other AI coding assistants. Let any MCP-compatible client benefit from existing `.claude/agents/` definitions, with zero additional configuration.

python

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "execveat-polyagent-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/execveat/polyagent-mcp",
        "polyagent-mcp"
      ],
      "env": {
        "POLYAGENT_MCP_DIR": ".claude/agents",
        "POLYAGENT_MCP_DEBUG": "true",
        "POLYAGENT_MCP_CONFIG": "<POLYAGENT_MCP_CONFIG>",
        "POLYAGENT_MCP_TIMEOUT": "900",
        "POLYAGENT_MCP_LOG_LEVEL": "INFO",
        "POLYAGENT_MCP_WORKSPACE": "/workspace"
      }
    }
  }
}

Polyagent MCP lets Claude Code agents run as separate CLI sessions within Codex, Gemini CLI, and other MCP-compatible clients. This enables zero-configuration reuse of existing .claude/agents/ definitions across different tools, delivering context-optimized agents with improved prompt adherence and performance.

How to use

After you configure an MCP client to connect to Polyagent MCP, you will automatically discover all agents defined in .claude/agents/. You can list available agents, review each agent’s instructions, and invoke agents by name. For example, call the tool named after an agent (such as commit-agent) and supply your task and context to perform the action.

How to install

Prerequisites: you need Python 3.10+ and at least one of the CLI tools for Codex, Claude, or Gemini installed on your system. You also need network access to fetch MCP components.

Install and run Polyagent MCP directly from source or via the standard install method.

# Quick Start with uvx (recommended)
uvx --from git+https://github.com/execveat/polyagent-mcp polyagent-mcp

# HTTP transport on a custom port
uvx --from git+https://github.com/execveat/polyagent-mcp polyagent-mcp --transport http --port 8000

For local development, clone the project and install dependencies, then run in development mode.

git clone https://github.com/execveat/polyagent-mcp.git
cd polyagent-mcp
uv sync

# Run locally with the console script
uv run polyagent-mcp

# Or use FastMCP for development with auto-reload
uv run fastmcp dev polyagent_mcp.server:mcp

Configuration

Configure your MCP client to connect to Polyagent MCP using the following patterns. You can choose a local stdio setup or a managed HTTP transport, depending on your environment.

Codex and Gemini CLI users add the following environment setup to enable a 15-minute minimum timeout for agent operations.

[mcp_servers.agents]
command = "uvx"
args = ["--from", "git+https://github.com/execveat/polyagent-mcp", "polyagent-mcp"]
tool_timeout_sec = 900

Environment variables

You can customize runtime behavior with environment variables. All listed options are optional.

POLYAGENT_MCP_DIR – directory with agent Markdown files (default: `.claude/agents`)
POLYAGENT_MCP_CONFIG – optional YAML/JSON file for runner/agent config overrides
POLYAGENT_MCP_WORKSPACE – working directory for agent execution
POLYAGENT_MCP_TIMEOUT – default timeout in seconds (default: 900)
POLYAGENT_MCP_LOG_LEVEL – Python log level (default: INFO)
POLYAGENT_MCP_DEBUG – enable FastMCP debug mode

Usage tips

Once configured, your MCP client will discover all agents in .claude/agents/. You can list them, review individual agent instructions, and invoke the appropriate agent by name. Remember to provide a meaningful context and task description to ensure reliable agent execution.

Requirements

Python 3.10+ and at least one of the following CLIs: Codex, Claude, or Gemini. An existing .claude/agents/ directory with agent definitions is required.

License

MIT license.

Available tools

commit-agent

Reviews changes and constructs a commit message, then commits with an appropriate message using the agent workflow.

Built by

VeilStrat

AI signals for GTM teams