Build a personal operating system where AI remembers your goals and context
Imagine moving past the need to explain your situation and goals to an AI every time, to a state where the AI already knows you and can start working immediately. LifeOS is a system that turns AI coding tools from simple utilities into personal secretaries that understand your goals and context. It continuously tracks and assists the process of moving from your current state to your ideal state.
v8.761.0: A system unified by a single loop
The recently released v8.761.0 version applies the concept of a 'single loop' across all levels of the system. It is designed so that outputs become tasks, tasks become projects, and projects move toward the final goal, ensuring work proceeds only after direction is confirmed. The release also adds an Assay feature that measures how well outputs align with the ideal state, and an Errata feature that records failures and runs correction experiments, giving the system a structure for self-improvement. Installation and upgrades are now handled with a single command, and 50 skills have been published.
How to make AI coding tools your own
LifeOS operates as a layer on top of an AI coding harness, turning the raw capabilities provided by the harness into something that belongs to the user. If the harness is the engine, LifeOS represents all the components that steer that engine according to the user's purpose. The system stores the user's goals, contacts, and preferences, and provides intelligent routing that automatically triggers the appropriate workflow when you say, 'Investigate this.' It also includes a self-improvement feature that modifies itself based on what it has learned, becoming better aligned with the user's working style over time.
Installation and recovery: AI sets it up directly
Installation involves pasting a prompt into an AI coding harness. You select one of the capable agents, such as Claude Code, Cursor, or Codex, and ask it to read the installation page address and perform the setup. If you prefer the terminal, a one-line command for Claude Code on macOS and Linux is also provided. During the installation process, the AI asks for the user's permission before making any changes. Even if the system is damaged, user customizations in the USER/ directory are never touched during upgrades or installations, and settings are handled via merging rather than overwriting to preserve existing configurations. Additionally, Git-based version control allows you to roll back to a previous state at any time.
Security responsibility lies with the user
LifeOS is fundamentally a personal AI harness for an individual, and it is constantly changing, with security features added or modified in each release. Therefore, the security of LifeOS components or the system as a whole is not guaranteed, and the responsibility for system security lies entirely with the user.
