Bolt - AI Orchestrator for Termux

Thin wrapper around Anchor Engine for Qwen Code on Termux/Android

What is Bolt?

Bolt is a pre-configured instance of Anchor Engine designed for:

• Qwen Code integration on Termux/Android
• Local model orchestration (coming soon: llama.cpp integration)
• Zero-config memory for AI agents

Architecture

┌─────────────────────────────────────────────────────┐
│                    Bolt (this repo)                  │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐  │
│  │ user_       │  │ local-data/ │  │ Config &    │  │
│  │ settings.   │  │ (inbox,     │  │ Scripts     │  │
│  │ json        │  │ distilled)  │  │             │  │
│  └─────────────┘  └─────────────┘  └─────────────┘  │
└───────────────────────┬─────────────────────────────┘
                        │ depends on
                        ▼
┌─────────────────────────────────────────────────────┐
│           @rbalchii/anchor-engine (npm)             │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐  │
│  │ Engine      │  │ MCP Server  │  │ CLI Tools   │  │
│  │ (core)      │  │             │  │             │  │
│  └─────────────┘  └─────────────┘  └─────────────┘  │
└─────────────────────────────────────────────────────┘

Quick Start

# Install dependencies (ignores native build scripts for Termux)
npm install --ignore-scripts

# Start the engine
npm start

# Engine will be available at http://localhost:3161

Configuration

Edit user_settings.json to customize:

• Server: Port, API key
• Watcher: Paths to watch for chat files
• LLM: Model settings (for future local model support)
• MCP: Model Context Protocol server settings

API Endpoints

All Anchor Engine endpoints are available:

Endpoint	Description
GET /health	Health check
GET /v1/agent/discover	Discover installed AI agents
POST /v1/agent/add	Add agent chat directory
GET /v1/ingest/status	Ingestion progress
GET /v1/graph/export	Export knowledge graph
GET /v1/memory/search	Semantic search
POST /v1/memory/ingest	Ingest content

Directory Structure

bolt/
├── package.json          # Depends on @rbalchii/anchor-engine
├── user_settings.json    # Qwen Code configuration
├── local-data/           # User data
│   ├── inbox/           # Files to ingest
│   ├── external-inbox/  # External sources
│   └── distilled/       # Distilled knowledge
└── README.md

Related Projects

• anchor-engine-node - Core engine (npm: @rbalchii/anchor-engine)
• Qwen Code - AI coding assistant

License

AGPL-3.0-only