Lode

A local-first desktop application for organizing, searching, and analyzing your AI conversation history.

• Import Conversations: Import conversations from OpenAI and Claude JSON exports
• Full-Text Search: Fast SQLite FTS5 search across all conversations
• Vector Search: Semantic search over your conversations (embedding-based). Build the index via Import or Settings.
• Chat: RAG chat over your data (OpenAI/Anthropic). Ask questions and get answers grounded in your imported conversations.
• Analytics: Track usage patterns, word counts, activity over time, and more
• Organization: Tag, bookmark, and annotate conversations with notes
• Find Tools: Extract code blocks, links, TODOs, questions, dates, decisions, and prompts
• Export: Export conversations in Markdown, CSV, or JSON formats
• Local-First: All data stored locally in SQLite - your conversations never leave your machine
• Desktop App: Native desktop application built with pywebview

• Python 3.8 or higher
• Windows (primary platform), macOS/Linux (should work but not fully tested)

1. Clone the repository

   git clone git@github.com:fox4snce/lode.git
   cd lode

2. Create a virtual environment

   python -m venv .venv

3. Activate the virtual environment

   Windows:

   .venv\Scripts\activate

   macOS/Linux:

   source .venv/bin/activate

4. Install dependencies

   pip install -r requirements.txt

5. Optional — Vector Search and Chat: If you want semantic search or RAG chat over your data, export the embedding model once (downloads from the internet, saves to vendor/ which is gitignored):

   python tools/export_embedder_onnx.py --model bge-small

   Without this step, Vector Search and the vectordb index job will fail with "Model not found at ... vendor/embedder_bge_small_v1_5". Full-text search and the rest of the app work without it.

Simply run:

python app/launcher.py

This will:

• Start the FastAPI backend server (default port 8000; change in Settings → Server)
• Open a desktop window with the application interface

1. Initialize Database: On first launch, you'll see the Welcome screen. Click "Initialize Database" to create the SQLite database and set up all necessary tables.
2. Import Conversations: After initialization, go to the Import screen and select your OpenAI or Claude JSON export file. The import process will add your conversations to the database.
3. Explore: Browse conversations, search, organize, and analyze your data

• Main Screen: Browse and view all your conversations (with in-app find)
• Search: Full-text search and Vector Search (semantic) across conversations
• Chat: Ask questions over your data (RAG with OpenAI or Anthropic)
• Analytics: View detailed analytics about your conversation usage
• Find Tools: Extract and browse code blocks, links, TODOs, and more
• Export: Export individual conversations or use bulk export features
• Settings: Server port, database integrity, deduplication, cleanup

lode/
├── app/                # Desktop launcher (pywebview)
├── backend/            # FastAPI backend (routes, chat, vectordb, db, config)
├── database/           # Database schema creation scripts
├── docs/               # Documentation (API, release process)
├── static/             # Static files (CSS, JS, images)
├── templates/          # Jinja2 HTML templates
├── tests/              # Test suite
├── tools/              # Build and utility scripts
└── importers/          # Conversation import modules

For development setup and API documentation, see README_DEV.md.

Run the test suite:

python tests/run_all_tests.py

See PACKAGING.md.

• Backend: FastAPI, SQLite with FTS5
• Frontend: Jinja2 templates, HTMX, vanilla JavaScript
• Desktop: pywebview
• Embeddings: ONNX Runtime (local embeddings for Vector Search)
• Chat: LiteLLM (OpenAI, Anthropic)

This project is licensed under the MIT License - see the LICENSE file for details.

For issues, questions, or contributions, please open an issue on GitHub.

Contact: support@simplychaos.org

Contributions are welcome! Please feel free to submit a Pull Request.