llm_fix

llm_fix + llx MCP

This example shows how to connect pyqual to a Dockerized llx MCP service.

What it does

• llx runs as an MCP SSE service in Docker.
• pyqual uses pyqual mcp-refactor as a pipeline stage.
• The helper first runs llx_analyze, then calls the aider MCP tool with a refactor prompt.
• The run result is saved to .pyqual/llx_mcp.json, which is also visible in pyqual status.

Files

• Dockerfile - builds an image containing both llx and pyqual
• docker-compose.yml - starts the MCP service on http://localhost:8000/sse
• pyqual.yaml - pipeline config that uses the MCP refactor stage

Quick start

1. Build and start the MCP service:

docker compose -f examples/llm_fix/docker-compose.yml up --build -d

2. Run the refactor workflow from your project directory:

pyqual mcp-refactor --workdir . --project-path /workspace/project

Or run the full pipeline with the refactor stage enabled:

export PYQUAL_LLX_MCP_URL=http://localhost:8000/sse
export PYQUAL_LLX_PROJECT_PATH=/workspace/project
pyqual run -c pyqual.yaml

3. Inspect the latest refactor run:

cat .pyqual/llx_mcp.json
pyqual status

Recommended setup

• Use PYQUAL_LLX_PROJECT_PATH=/workspace/project so the host project maps to the same path inside the MCP container.
• Keep PYQUAL_LLX_USE_DOCKER=false unless you want the aider tool to spawn its own nested Docker container.
• If your llx routing relies on Ollama or another backend, expose that backend to the container as well.
• For local development, you can also start the persistent service with pyqual mcp-service --host 0.0.0.0 --port 8000.

Plugin workflow

You can register the plugin with:

pyqual plugin add llx-mcp-fixer

That will append a ready-to-customize llx-mcp-fixer snippet to pyqual.yaml.