How to Deploy OpenClacky with Docker: Step-by-Step Tutorial

In this tutorial you will deploy OpenClacky — the token-efficient, MIT-licensed AI agent — using the official GHCR Docker image, connect it to an OpenAI-compatible model, and run your first agentic task. Total time: about 10 minutes.

Prerequisites

• Docker Engine 24+ (or Docker Desktop) with the compose plugin
• An API key for any OpenAI-compatible provider (OpenAI, DeepSeek, or a relay)
• At least 2 vCPU and 4 GB RAM (4 vCPU / 8 GB recommended for heavy sessions)

Step 1 — Pull the image

docker pull ghcr.io/clacky-ai/openclacky:latest

Images are published to GitHub Container Registry on version tags. Pinning to a release tag instead of latest is recommended for production.

Step 2 — Run with docker compose

Save this docker-compose.yml and run docker compose up -d:

services:
  openclacky:
    image: ghcr.io/clacky-ai/openclacky:latest
    restart: unless-stopped
    ports:
      - 8080:8080
    volumes:
      - ./data/openclacky:/data

Step 3 — Configure your model key

OpenClacky is BYOK: set your OpenAI-compatible endpoint and key, then choose the model routing strategy:

Step 4 — Run your first task

1. Start a session — sessions never restart, so cache hits stay near 100%.
2. Invoke a Skill with / — browse, fuzzy-search, or call directly.
3. Create a new Skill in natural language; the agent drafts SKILL.md, validates it, and self-evolves it after each run.

Verification checklist