stackship 0.1.1

Deploy AI stacks with one command. Available: openclaw (OpenClaw+Ollama+Qdrant), ollama, rag. Local Docker Compose or cloud (Pro) with Terraform.

🚀 StackShip

Deploy AI stacks with one command.

StackShip takes the pain out of setting up AI infrastructure. One command gets you a production-ready stack — locally with Docker Compose, or in the cloud with Terraform.

Overview

StackShip is a Python CLI that deploys pre-built AI stacks so you can run OpenClaw, Ollama, Qdrant, and similar tools without wiring Docker and config by hand.

• Local: Docker Compose stacks under ~/.stackship/deployments/<stack>/
• Cloud (Pro): Terraform-based deploys (Azure supported; AWS, GCP, DigitalOcean planned)

Tech stack: Typer + Rich for the CLI, Jinja2 for templates, docker compose (with fallback to standalone docker-compose). Cloud: Terraform with Azure provider. Python 3.9+.

How it works: Templates live in stackship/templates/ (e.g. openclaw, ollama, rag). Each has a docker-compose.yml.j2, optional config/ and env.example. The deploy flow creates the deployment dir, renders templates, runs docker compose pull and up, and can pre-pull Ollama models. Use stackship doctor to check Docker, disk, and ports.

Highlights: Clear separation (CLI, deploy engine, templates, doctor), good UX (progress, dry-run, panels), CI on Python 3.10–3.12 with ruff and pytest. OpenClaw stack is the primary, fully supported path.

Current limitations: Pro/cloud path requires a license key (check not yet implemented). Cloud deploy exits after the license prompt — Terraform is not invoked from the CLI yet. The ollama and rag templates are in the registry but may be incomplete.

Quick Start

pip install stackship

# Deploy OpenClaw stack locally (free)
stackship deploy openclaw
# Gateway: http://localhost:18789   Qdrant: http://localhost:6333

# Deploy to Azure (pro)
stackship deploy openclaw --cloud azure

The OpenClaw stack uses Ollama as the default model (ollama/llama3.2). You can optionally provide an Anthropic API key during setup for Claude, or add it later to ~/.stackship/deployments/openclaw/.env.

What You Get

stackship deploy openclaw

A complete AI assistant stack:

• OpenClaw — personal AI agent framework (gateway on port 18789 by default)
• Ollama — local LLM inference (Llama, Mistral, Qwen, etc.); default model is ollama/llama3.2
• Qdrant — vector database for memory (API and dashboard on port 6333)

All pre-configured, networked, and ready to go. Deployment files and secrets live in ~/.stackship/deployments/openclaw/ (including .env and config/). To change ports or re-run compose, use the same directory:

cd ~/.stackship/deployments/openclaw
# Optional: edit docker-compose.yml or .env, then:
docker compose up -d

Pricing

Tier	What	Price
Free	Local Docker Compose stacks	$0
Pro	Cloud templates (Terraform), priority support	$15/mo or $29 one-time

Stack Templates

Available Now

• openclaw — Full OpenClaw + Ollama + Qdrant
• ollama — Standalone Ollama with model management
• rag — RAG pipeline (Qdrant + embeddings API)

Coming Soon

• langchain — LangChain + vector store + API
• comfyui — ComfyUI + model cache

Commands

stackship deploy <stack>          # Deploy a stack locally
stackship deploy <stack> --cloud <provider>  # Deploy to cloud (pro)
stackship list                    # List available stacks
stackship status [<stack>]         # Check running stacks
stackship stop <stack>            # Stop a stack (preserves data)
stackship destroy <stack>         # Remove a stack (use --volumes to remove data)
stackship logs <stack>            # View stack logs
stackship doctor                  # Diagnose common issues

Deploy options (e.g. for openclaw): --port / -p (gateway port), --ollama-port (expose Ollama on host), --dry-run (render config only), --gpu, --models llama3.1,mistral, --api-key / -k (Anthropic), --skip-setup (skip interactive prompts).

Cloud Providers (Pro)

Provider	Status
Azure	🔜 Coming Soon
AWS	🔜 Coming Soon
GCP	🔜 Coming Soon

Requirements

• Python 3.9+
• Docker & Docker Compose (for local deploys)
• Terraform (for cloud deploys, auto-installed if missing)

License

MIT — Free and open source.

Cloud templates require a Pro license key.

---

Built by ai-engineering-lab · Dare to Dream, Inspire Leadership, and Spark Innovation Through Diverse Ideas.