home / mcp / enhanced poe mcp server

Enhanced POE MCP Server

Provides an integrated POE MCP server with Warp.dev access, context-aware queries, streaming responses, and production-grade reliability.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "anansitrading-poe-proxy-mcp": {
      "url": "https://your-app.up.railway.app/mcp/stream",
      "headers": {
        "DEBUG_MODE": "false",
        "MAX_RETRIES": "3",
        "POE_API_KEY": "your_poe_api_key_here",
        "OPENAI_API_KEY": "your_openai_key",
        "RATE_LIMIT_RPM": "500",
        "METRICS_ENABLED": "true",
        "MAX_FILE_SIZE_MB": "10",
        "STREAM_BUFFER_SIZE": "64",
        "HEALTH_CHECK_INTERVAL": "30",
        "USE_CLAUDE_COMPATIBLE": "true",
        "SESSION_EXPIRY_MINUTES": "60"
      }
    }
  }
}

You install and run an MCP server that bridges POE models with Warp.dev, giving you context-aware, streaming AI assistance inside your terminal. It supports multiple models, safe command execution, file operations, session management, and health/metrics monitoring, so you can build automated workflows and real-time AI help around your terminal tasks.

How to use

Start by connecting to the MCP server from your Warp terminal client. You will interact with POE models through Warp using natural queries and terminal context. Use context-aware queries to have the model consider your current working directory, terminal output, and git state. You can stream long responses in real time, execute safe terminal commands, and perform file operations directly from model responses. Manage your session to maintain conversation context across multiple queries. When you need a quick health or metrics snapshot, request a health check or metrics readout.

Practical usage patterns include: getting code suggestions while editing files, reviewing configuration files, or querying model knowledge about your current project setup. You can specify models from POE such as GPT-4o or Claude 3 Sonnet, and you can enable streaming to see code or guidance as it’s generated. Remember to clear the session when you want a fresh context.

How to install

Prerequisites: you need Linux, macOS, or WSL2, Python 3.8 or higher, Warp Terminal, and a POE API key.

Step 1: Clone and set up the project

# Clone the repository
git clone https://github.com/Anansitrading/enhanced-poe-mcp.git
cd enhanced-poe-mcp

# Quick installation with script
chmod +x install.sh
./install.sh

# Or manual setup
python3 -m venv venv
source venv/bin/activate  # On Linux/macOS
pip install -e .

Step 2: Configure environment variables

# Create a .env file with credentials
POE_API_KEY=your_poe_api_key_here
OPENAI_API_KEY=your_openai_key  # Optional for OpenAI SDK compatibility
DEBUG_MODE=false
USE_CLAUDE_COMPATIBLE=true
MAX_FILE_SIZE_MB=10
SESSION_EXPIRY_MINUTES=60
RATE_LIMIT_RPM=500
MAX_RETRIES=3
STREAM_BUFFER_SIZE=64
METRICS_ENABLED=true
HEALTH_CHECK_INTERVAL=30

Step 3: Configure in Warp Terminal

{
  "enhanced-poe-mcp": {
    "command": "python3",
    "args": ["/home/david/Projects/Kijko/MVP/MVP_Kijko/enhanced-poe-mcp/poe_server_phase2.py"],
    "env": {
      "POE_API_KEY": "your_poe_api_key_here",
      "PYTHONUNBUFFERED": "1",
      "DEBUG_MODE": "false",
      "RATE_LIMIT_RPM": "500"
    }
  }
}

Additional sections

Server variants and how to run them are described in the setup guide. You can also deploy remotely to keep the service available even when your local machine is off. The production server includes health checks, metrics, and rate limiting to keep things stable in production.

Configuration and security notes: keep your POE API key secure, set appropriate rate limits, and ensure that environment variables are loaded in your deployment environment. Use the remote hosting option for stability in production.

Troubleshooting quick tips: ensure Python paths are correct, verify file permissions, test the server manually, and consult logs for errors. If authentication fails, re-check your POE_API_KEY and its placement in the environment.

Remote deployment and hosting

For high availability, host your MCP server remotely. Use Railway for a straightforward deployment or choose Render, Google Cloud Run, Fly.io, or similar services. The remote server URL will be used by Warp to connect to the MCP endpoint.

Example remote setup using Railway

web: python poe_server_phase2.py

Configure Warp to point at the remote URL: use the remote URL provided by your hosting service, for example https://your-app.up.railway.app/mcp/stream, and ensure you set the POE_API_KEY and other necessary environment variables in the hosting service.

Monitoring & troubleshooting

Health checks expose the status of the server and help you detect issues early. Metrics provide visibility into throughput and latency. Use the health and metrics endpoints to monitor the server's health in production.

Available tools

ask_poe

Basic POE query with a model and prompt

ask_with_attachment

Query with a file attachment for analysis by a POE model

ask_poe_with_warp_context

Context-aware query that considers Warp context

stream_poe_to_warp

Streaming response endpoint for long-running queries

execute_warp_action

Run terminal actions or commands with validation

clear_session

Reset conversation context for a fresh start

list_available_models

List all available POE models

health_check

Check server health status

get_metrics

Retrieve performance metrics