Self-hosted focus app with AI companion and session tracking
focus-with, from Sayhi2e10414 Cpu, is a self-hosted productivity ecosystem that converts vague plans into finished work. The app centers on a single-task session loop and optional AI companions using the Model Context Protocol (MCP). It bundles timed sessions, session recovery after distraction, and cross-device continuity while supporting Markdown plan imports. Designed for developers, knowledge workers, and AI power users who need local control and AI integration during focused work.
What tasks the tool actually handles and how it fits a workflow
The app maps planning to execution by letting users turn goals into directions, projects, and actionable tasks, with Markdown plan import supported. Its core workflow asks you to pick one task, start a session, return when distracted, then decide what comes next. This single-task loop targets work that benefits from discrete, time-boxed effort and repeated short sessions for progress tracking.
How AI companions participate in real sessions and what they can do
AI assistants can act on your focus state because the server exposes annotated tools to models. The app provides 11 annotated MCP tools that allow compatible clients to start sessions, query focus status, and manage tasks. AI features are optional and the core server operates without an AI API key, so assistants augment rather than replace the local session controls.
Installation, platforms, and technical limits for deployment
Self-hosting requires minimal developer setup: the server runs on Python 3.11+ or via Docker and presents a web interface plus native front-ends. Native macOS has a floating timer, Windows uses an experimental WinUI 3 timer, Android has a companion app, and iOS integrates through Shortcuts. Cross-device state sync covers web, desktop, and mobile endpoints when configured.
Privacy model and distraction handling in practice
Local-first defaults and opt-in remote features mean installations bind toand data stays on your server unless you enable remote MCP options. Distraction handling ties sessions to phone events and offers optional camera-based detection for nudges and consequences. The app also records daily and weekly analytics, merging repeated sessions on the same task into unified totals for progress review.
Practical choice for technically minded users who want local AI-aware focus
The app rewards users who prioritize data locality and programmatic assistant control, and it suits developers or power users who can run a local server. Setup and the experimental Windows timer create a learning curve for non-technical users, so expect configuration work before it becomes part of a daily routine. For those comfortable with self-hosting, it provides a privacy-centered bridge from planning to tracked work.
