Walkthroughs

Three short, looping walkthroughs of the core loops. Each has a video script (storyboard + narration + copy-paste commands) you can re-record from, and an animated GIF captured against a live stack.

The images below are the poster frames — real screenshots captured from a live dashboard by scripts/capture_docs_media.mjs (Playwright). The animated GIF version of each (media/img/<scene>.gif) is produced by the same harness with a terminal recording spliced in per each script's Capture notes; run a capture and it swaps in. See docs/media/README.md.

Quickstart — running in 5 minutes

One CLI, five commands, a full self-hosted stack on localhost.

Quickstart walkthrough — install to running dashboard

Script: quickstart.md
Page: Quickstart · Hands-on: TRY_MYAI.md

The Brain — git-versioned agent memory

Memory is a real git repo: sessions are branches, wrap-up is a merge. What one agent learns, every future agent boots from.

Brain walkthrough — init, session branch, stash/pop, merge, boot from brief

Script: brain.md
Page: Concepts → Memory tiers · Hands-on: TRY_BRAIN.md

Continuity — kill an agent, another picks it up

Your context lives in your layer, not the agent's. Kill a session mid-task and any blank agent plugs in already knowing who you are and what's next.

Continuity walkthrough — kill agent A, blank agent B continues

Script: continuity.md
Page: Concepts → The continuity model · Full demo: CONTINUITY_DEMO.md

Next: read the Concepts page for the model behind these loops, or the Quickstart to run them yourself.