StackForge

CPU-only, privacy-focused Docker Compose stack for AI agents — local LLM, vector search, long-term memory, and synced notes, all self-hosted

What This Is

StackForge bundles the pieces an AI agent needs to run privately on your own hardware: a local LLM, a vector database, persistent memory, private search, a synced notes vault, and browser automation. It targets CPU-only boxes — a VPS, homelab server, or bare-metal machine — and keeps all agent context on infrastructure you control instead of a vendor's.

Built for privacy-conscious developers, homelab operators, and anyone who wants a complete AI agent stack without sending data to the cloud.

Quick Start

Docker Compose (recommended)

git clone https://github.com/OneByJorah/StackForge.git
cd StackForge
cp .env.example .env
docker compose up -d

Open http://localhost:8083 for the vault viewer.

For an interactive first-run wizard that generates secrets and sets your IP, run bash bootstrap.sh instead.

pip

pip install stackforge

From Source

git clone https://github.com/OneByJorah/StackForge.git
cd StackForge
pip install -r requirements.txt
python3 -m stackforge

Install

Docker Compose

git clone https://github.com/OneByJorah/StackForge.git
cd StackForge
cp .env.example .env
docker compose up -d

pip

pip install stackforge

From Source

git clone https://github.com/OneByJorah/StackForge.git
cd StackForge
pip install -r requirements.txt
python3 -m stackforge

Features

• One-command deploy — docker compose up -d (or bootstrap.sh for guided setup)
• Interactive first run — bootstrap.sh prompts for secrets, generates random passwords, seeds the vault
• Local LLM — Ollama-based, CPU-only inference, no GPU required
• Vector database — ChromaDB for long-term memory and semantic search
• Persistent memory — Honcho-backed long-term memory store
• Private search — Self-hosted SearXNG for private web search
• Synced notes vault — Obsidian-compatible notes synchronized across agents
• Browser automation — Playwright-powered browser for web automation
• CPU-only — No GPU required, runs on any VPS or homelab server
• Privacy-first — All agent context stays on your hardware

Tech Stack

• Orchestration — Docker Compose (14 services)
• LLM — Ollama (CPU inference)
• Vector DB — ChromaDB
• Memory — Honcho
• Search — SearXNG
• Notes — Obsidian vault
• Browser — Playwright
• Python — 3.11+, pydantic, pyyaml, requests, chromadb, numpy

Package Badges

Architecture

┌─────────────────────────────────────────────────────────┐
│                   StackForge Stack                       │
│  (14 services, CPU-only, Docker Compose)                │
│                                                         │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌────────┐ │
│  │  Ollama  │  │ ChromaDB │  │ Honcho   │  │SearXNG │ │
│  │  (LLM)   │  │(Vector)  │  │ (Memory) │  │(Search)│ │
│  └──────────┘  └──────────┘  └──────────┘  └────────┘ │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌────────┐ │
│  │  Vault   │  │  Nginx   │  │ Playwright│  │ Redis  │ │
│  │(Notes)   │  │(Proxy)   │  │ (Browser) │  │(Cache) │ │
│  └──────────┘  └──────────┘  └──────────┘  └────────┘ │
└─────────────────────────────────────────────────────────┘

Configuration

Copy .env.example to .env and configure:

Variable	Description
OLLAMA_MODEL	Default Ollama model (default: llama3.2)
VAULT_PORT	Vault viewer port (default: 8083)
SEARXNG_URL	SearXNG instance URL
REDIS_URL	Redis connection URL

For interactive setup: run bash bootstrap.sh — generates secrets, sets IP, seeds vault.

Services

Service	Purpose
ollama	Local LLM inference (CPU)
chromadb	Vector database for memory
honcho	Long-term memory store
searxng	Private web search
vault	Obsidian notes viewer (port 8083)
nginx	Reverse proxy / gateway
playwright	Browser automation
redis	Cache and session store
+ 6 more supporting services

Contributing

Contributions are welcome. Open an issue to report a bug or request a feature.

License

MIT — see LICENSE.