OpenHands + LM Studio Setup

Local autonomous AI agent setup using OpenHands (Docker) and LM Studio (LLM Provider).

Quick Start

1. LM Studio:
   • Load a model (Recommended: qwen2.5-coder-7b-instruct or similar).
   • Start Local Server on port 1234.
2. Start OpenHands:

   docker-compose up -d

3. Access: Open http://localhost:3000.

Configuration

Settings are in .env.

• LLM_MODEL: Must match the model ID loaded in LM Studio (e.g., openai/qwen/qwen3-coder-30b).
• LLM_BASE_URL: http://host.docker.internal:1234/v1 (Required for Docker -> Host communication).

To switch models:

1. Edit .env (uncomment desired model).
2. Restart: docker-compose restart.

Troubleshooting

Run the pre-flight check to verify connectivity:

./preflight-check.sh

File Structure

• docker-compose.yml: Container definition.
• .env: Configuration variables.
• workspace/: Shared directory for agent files.