Sovereign Dark Factory (local-dark-factory)

A local-first, zero-cloud-token AI Software Factory that coordinates coding agents in fast, isolated git worktrees, enforces deterministic verification gates, preserves patches and execution evidence, and integrates human-in-the-loop governance.

"Make sure you own your AI. AI in the cloud is not aligned with you; it’s aligned with the company that owns it."

📖 AI-Native SDLC Architecture

This project follows the AI-Native SDLC Playbook:

🔗 The Local Stack Integration

Unlike existing cloud-bound prototypes, the Sovereign Dark Factory runs entirely on your workstation without cloud dependencies:

Task Submission (CLI / Trigger)
       │
       ▼
Durable SQLite Workflow Engine (.factory/factory.db)
       │
       ▼
Disposable Git Worktree Sandbox
       │
       ▼
Local Agent Harness (local-coder)
       ├── Ollama (:11434): qwen2.5-coder:14b / 7b
       └── Prism CUDA (:5272): phi4-mini / ONNX GenAI (Prism)
       │
       ▼
Deterministic Verification Gates (test.sh / pytest / cargo / npm)
       │
       ├──[Failed]──► AST Self-Healing Loop (max N retries)
       │
       └──[Passed]──► Evidence Locker (.factory/runs/<RUN_ID>/diff.patch)
                            │
                            ▼
               Human-in-the-Loop Review Gate (dark-factory review --approve)

🚀 Quick Start (Development)

# 1. Health check local engines & GPU
python3 -m dark_factory.cli doctor

# 2. Run unit tests
pytest tests/ -v

# 3. Submit a dark factory task
python3 -m dark_factory.cli run \
  --repo /path/to/repo \
  --task "Implement thread-safe token bucket limiter" \
  --gates default

# 4. Run the repeatable real-model benchmark suite
dark-factory eval --list

# 5. Browse run history, evidence, and eval reports (local, read-only, 127.0.0.1 only)
dark-factory dashboard