LifeOS is an AI-powered Life Operating System built in TypeScript that captures personal goals, values, and context, then uses AI agents to help users move from their current state toward their ideal state through verifiable steps. The system is designed to run on top of capable AI coding harnesses like Claude Code, Cursor, or other agentic platforms, functioning as a persistent layer that adds personalization and structure to raw AI capabilities.
The installation process is distinctive in that LifeOS installs itself through an AI agent. Users paste an installation prompt into their chosen AI harness, and the agent handles the entire setup process, asking for permission before making changes. A terminal shortcut is available for Claude Code users on macOS and Linux. The system requires a capable AI coding harness and bun as dependencies.
Core functionality centers on five key differentiators from using an AI harness alone. LifeOS provides persistent memory across sessions, allowing the AI to remember past decisions and learnings. It includes custom skills—specialized capabilities bundled into a single self-contained skill that covers research, security, writing, art, and more. The system maintains user context including goals, contacts, and preferences without requiring re-explanation. Intelligent routing automatically triggers appropriate workflows based on user requests. Finally, the system self-improves by modifying itself based on learned patterns.
The repository is harness-agnostic by design, built on universal primitives like hooks, skills, context files, and agentic routing rather than vendor-specific features. While Daniel Miessler builds and runs it on Claude Code, the TypeScript and Bash codebase is designed to port to any capable agent. The system distinguishes itself from Miessler's earlier Fabric project, which is a collection of AI prompts for specific tasks. LifeOS instead provides infrastructure for how an AI assistant operates, including memory management, skill routing, context handling, and self-improvement mechanisms.
Recovery and safety are built into the design. Users can back up their configuration before upgrades, and customizations in the USER directory are never touched by installers or upgrades. Settings merge rather than overwrite, version control preserves history, and the AI assistant can help repair issues. Re-running the installer detects existing installations and merges intelligently.
The roadmap includes local model support for privacy and cost control through Ollama and llama.cpp, granular model routing to direct different tasks to appropriate models, remote access across devices, outbound phone calling capabilities, and external notifications via email, Discord, Telegram, and Slack.
GitGenius activity data shows the repository has 601 tracked issues and pull requests with a median response latency of 17.7 hours and mean latency of 141.8 hours. Daniel Miessler leads contribution activity with 794 tracked events, followed by kaimagnus with 286 events and xbt-a4224j with 34 events. The repository connects to related projects including danielmiessler/personal_ai_infrastructure, anthropics/claude-code, and conda/conda through overlapping contributors. The project is classified across multiple domains including Personal AI, Local AI, AI Infrastructure, Machine Learning, LLM Setup, AI Environment, Data Privacy, AI Tools, Automation, and Self-hosted AI. The system is released under the MIT License and accepts community contributions through GitHub Issues and pull requests.
