Run Codex, Claude Code, and Hermes Locally with a Unified Docker Container

Description and use

Developers can run Codex, Claude Code, and Hermes locally in a single Docker container using a unified protocol, simplifying setup and removing cloud dependencies.

Problem

Developers need to run multiple LLM agents (Codex, Claude Code, Hermes) on their own machines without relying on cloud APIs.

Outcome

Developers can run Codex, Claude Code, and Hermes locally in a single container, simplifying setup and eliminating cloud dependencies.

Workflow

• 1. Clone the HarnessRouter repository: git clone https://github.com/HarnessRouter/harnessrouter
• 2. Build or pull the Docker image: docker pull harnessrouter/harnessrouter:latest
• 3. Configure environment variables (e.g., model paths, API keys if needed).
• 4. Run the container: docker run -p 8000:8000 -e MODEL_PATH=/models -v /local/models:/models harnessrouter/harnessrouter:latest
• 5. Send requests to the unified endpoint (e.g., http://localhost:8000/run) using the same payload format for any agent.

Evidence

Runs agents like Codex, Claude Code, and Hermes on your own machine using a single Docker container and unified protocol.

Source