What it does FreeTodo is a desktop todo app (Electron + FastAPI backend) that wraps a familiar task manager—hierarchical tasks, drag-and-drop calendar, tags, priorities—in an AI chat interface. You describe what you want to do; it asks clarifying questions and spits out subtasks. It also extracts action items from chat history and suggests next steps based on your existing todo context.

The interesting bit The project is explicitly framed as “context engineering,” not just another LLM wrapper. It stores your activity traces (screenshots, eventually audio) in ChromaDB alongside tasks, with the stated goal of building enough personal context that the system can eventually offer proactive suggestions—like a secretary who has actually been paying attention. The roadmap is refreshingly honest about what’s vaporware and what ships today.

Key highlights

• AI task breakdown via guided questionnaire flow (working now)
• Multi-panel UI: todo list, chat, and detail view in one draggable layout
• Calendar with day/week/month views and drag-to-schedule
• Social software integrations: WeChat and Feishu message extraction already implemented
• Voice input, multimodal capture, and “AI secretary personification” listed as future work
• Full English/Chinese i18n support

Caveats

• v0.1 status: many headline features (voice input, proactive services, local LLM optimization) are roadmap items, not shipped
• Requires running both a Python 3.12 backend and a Next.js frontend; setup is manual, not one-click despite install scripts being commented out
• The “Dynamic Island” UI element and agent tool-calling are marked as in-progress

Verdict Worth a look if you’re building or researching personal AI assistants and want to see an open, explicit roadmap for context-aware task management. Skip it if you need a polished, batteries-included productivity app today—this is a research vehicle with a todo interface glued on top.