Skip to content

ContainerAPP.md

Latest commit

 

History

History
87 lines (62 loc) · 2.4 KB

ContainerAPP.md

File metadata and controls

87 lines (62 loc) · 2.4 KB

MagesticAI Docker Container - Test & Deploy Steps

Build & Run

cd <your-clone-dir>          # the directory you cloned MagesticAI into

# Build and start (clean)
sudo docker compose down -v && sudo docker compose build && sudo docker compose up -d

# Start (no rebuild)
sudo docker compose up -d

Access

  • URL: http://${INSTANCE_IP}:3101 (from .env) — or http://localhost:3101 if not using macvlan
  • Token: Auto-generated on first run, retrieve with:
sudo docker exec magesticai cat /home/magesticai/.magestic-ai/.token

Useful Commands

# Check container status
sudo docker compose ps

# View logs (last 30 lines)
sudo docker logs magesticai --tail 30

# Follow logs in real time
sudo docker logs magesticai -f

# Shell into container (as magesticai user)
sudo docker exec -it magesticai bash

# Shell as root
sudo docker exec -it -u root magesticai bash

# Check Claude Code CLI inside container
sudo docker exec magesticai bash -l -c "claude --version"

Stop & Clean Up

# Stop container (keeps volumes)
sudo docker compose down

# Stop and remove volumes (full reset)
sudo docker compose down -v

# Remove image too
sudo docker compose down -v --rmi all

Environment Variables

Set in docker-compose.yml or .env file. Key vars:

Variable Default Description
APP_HOST 0.0.0.0 Listen address
APP_PORT 3101 Server port
APP_API_TOKEN (auto-generated) Auth token for login
APP_DEBUG false Enable Swagger docs at /docs
APP_DEFAULT_SHELL /bin/bash Default terminal shell
APP_MAX_TERMINALS 20 Max concurrent terminals

Architecture

  • Base image: Ubuntu 24.04
  • Runtime user: magesticai (non-root)
  • Python venv: /home/projects/MagesticAI/.venv
  • Node.js: Copied from build stage (for frontend build + npm available at runtime)
  • Frontend: Pre-built static files served from apps/web-server/static/
  • Data directory: /home/magesticai/.magestic-ai/ (persisted via Docker volume)

Onboarding Flow

  1. Login with token
  2. Onboarding wizard launches automatically
  3. Install Claude Code CLI (installs Node.js via fnm if needed, then npm install -g @anthropic-ai/claude-code)
  4. Configure OAuth: runs claude setup-token in embedded terminal, auto-detects token from output
  5. Add a project (default browse path: /home)