CodeForge AI

Local-first, privacy-first AI coding assistant.

Quick Start

# Create environment
conda env create -f environment.yml
conda activate codeforge-ai

# Install in development mode
pip install -e ".[dev]"

# Run hardware detection
codeforge hardware

# Run diagnostics
codeforge doctor

# Start API server
codeforge serve

Architecture

VS Code Extension
        |
        v
CodeForge AI Client
        |
        v
CodeForge AI Server (FastAPI)
        |
   +----+----+
   |         |
   v         v
Agent     Model
Engine    Runtime

Development Phases

Completed

• Phase 1: Hardware Detection
• Phase 2: Runtime Abstraction
• Phase 3: Model Management
• Phase 4: Text Generation
• Phase 5: FastAPI Server
• Phase 6: VS Code Extension
• Phase 7: Repository Intelligence
• Phase 8: Hybrid Search + Embeddings + RAG
• Phase 9: Secure Tool System

Remaining

• Phase 10: File Editing + Terminal
• Phase 11: Agent Engine
• Phase 12: Memory
• Phase 13: Security Hardening
• Phase 14: Cross-platform Packaging
• Phase 15: Performance Optimization

Project Structure

CodeForge-AI/
  codeforge/           # Main Python package
    packages/          # Core packages
      hardware/        # Phase 1: Hardware detection
      runtime/         # Phase 2: Runtime abstraction
      models/          # Phase 3: Model management
      generation/      # Phase 4: Text generation
      indexing/        # Phase 7: Repository indexing
      embeddings/      # Phase 8: Embedding providers
      chunks/          # Phase 8: Code chunking
      vector/          # Phase 8: Vector store
      search/          # Phase 8: Search engine
      rag/             # Phase 8: RAG pipeline
    api/               # FastAPI server (Phase 5)
    cli.py             # CLI entry point
  codeforge-vscode/    # VS Code extension (Phase 6)
  tests/               # Test suite
  docs/                # Documentation

Testing

# Run all tests (515+ tests)
pytest tests/ -v

# Run specific phase tests
pytest tests/unit/test_search_rag.py -v    # Phase 8: Search + RAG
pytest tests/unit/test_indexing.py -v       # Phase 7: Repository Intelligence
pytest tests/unit/test_api.py -v           # Phase 5: API Server

# Linting
ruff check codeforge/ tests/

CLI Commands

# Hardware & Diagnostics
codeforge hardware          # Show detected hardware
codeforge doctor            # Run diagnostics

# Runtime
codeforge runtime           # Show runtime info
codeforge device            # List available devices

# Model Management
codeforge models            # List registered models
codeforge model list        # List models
codeforge model inspect     # Inspect model details
codeforge model load        # Load a model
codeforge model unload      # Unload a model
codeforge model discover    # Discover new models

# Generation
codeforge generate          # Generate text
codeforge chat              # Interactive chat

# Search & RAG (Phase 8)
codeforge search "UserService"    # Search code
codeforge rag "How does auth work?"  # RAG query
codeforge embeddings-index        # Build embeddings
codeforge embeddings-status       # Check embedding status

# Server
codeforge serve              # Start API server

API Endpoints

Health

• GET /v1/health — Health check
• GET /v1/health/ready — Readiness check

Hardware

• GET /v1/hardware — Hardware info
• GET /v1/hardware/gpu — GPU info

Runtime

• GET /v1/runtime — Runtime info
• GET /v1/runtime/capabilities — Available capabilities

Models

• GET /v1/models — List models
• GET /v1/models/{id} — Get model
• POST /v1/models/{id}/load — Load model
• POST /v1/models/{id}/unload — Unload model

Generation

• POST /v1/generate — Generate text
• POST /v1/generate/stream — Streaming generation

Embeddings

• POST /v1/embeddings — Create embedding
• POST /v1/embeddings/batch — Batch embeddings
• GET /v1/embeddings/health — Embedding health
• GET /v1/embeddings/config — Embedding config

Search

• POST /v1/search — Search code
• GET /v1/search/modes — Available search modes

RAG

• POST /v1/rag/query — RAG query

Workspaces

• GET /v1/workspaces — List workspaces
• POST /v1/workspaces/index — Index workspace
• GET /v1/workspaces/{id} — Get workspace
• GET /v1/workspaces/{id}/files — Get files
• GET /v1/workspaces/{id}/symbols — Get symbols
• GET /v1/workspaces/{id}/dependencies — Get dependencies
• POST /v1/workspaces/{id}/refresh — Refresh index

OpenAI Compatible

• POST /v1/chat/completions — Chat completions
• POST /v1/completions — Text completions

License

MIT