digital-twinPersonal Agent OS Blueprint

Build an AI operating layer that retrieves your knowledge, routes to your skills, writes durable outputs, and learns back into your workflow.

5-Minute Twin Demo

Inspect the public sample files, choose a capability, run one prompt, and verify the durable write-back loop.

New to the repo? Start with a short explanation of the main terms and the files that prove each one.

Reads wiki/, prior outputs, summaries, and agent-learnings/ before acting, so context lives outside the chat window.

Classifies the request and chooses the matching capability instead of sending every task through one generic prompt.

Keeps writing, research, wiki management, resume work, site improvement, and learning loops as modular workflow files.

Produces durable files and saves reusable rules, preferences, and failure modes for the next run.

Shows when an agent can auto-run, when it needs review, and when it must stop before touching risky work.

Gives reviewers a simple template for checking what was requested, which files changed, what evidence exists, and what still needs human review.