Lyn is a privacy-first, local/hybrid AI assistant designed for both daily users and power users. It prioritizes local operation, data privacy, and user control.
- Local-First: Operate primarily locally.
- Privacy: Strong focus on user data privacy and anonymity.
- Hybrid LLM: Support local and web-based LLMs.
- Task Automation: Automate various digital tasks.
- Summarization Memory: Use summarization for context, not chat history.
- Extensible Tools: Allow adding new capabilities via a tool system.
- Multi-Client: Offer GUI, Web, and CLI/TUI interfaces.
- Open Source: Primarily open-source codebase.
- Core Foundation: Project setup, async runtime, config, logging, local LLM interface (Ollama).
- Memory System: Qdrant integration, summarization logic, context retrieval.
- Tools Framework: Basic tools (calculator, date, web search).
- CLI Interface: Initial command-line client.
- Web LLM Integration: Server-mediated, anonymized access.
- Advanced Tools: File management, application installation.
- GUI/Web Clients: Develop graphical and web interfaces.
- Internationalization: Add multi-language support.
(More detailed roadmap tracked via project issues/epics)
Please see CONTRIBUTING.md for details on how to contribute to the project.
Lyn is licensed under either of the MIT License or the Apache License, Version 2.0 (See LICENSE-MIT and LICENSE-APACHE) at your option.