Skip to content

quickstart.md

Latest commit

 

History

History
47 lines (32 loc) · 2.67 KB

quickstart.md

File metadata and controls

47 lines (32 loc) · 2.67 KB
Raw

Quickstart

CapstanAI LabNote has three stages: acquire a copy, bootstrap the workspace, then operate it.

1. Acquire LabNote

First decide whether you want LabNote locally, on GitHub, or both. If you are already talking to an AI, the simplest starting point is:

Help me set up CapstanAI LabNote for this project. First work out what access you have, then recommend the simplest safe route for me. I want you to explain any terminal commands before I run them.

Preferred routes:

  • Browser AI + terminal available: AI-guided local clone, with a private GitHub origin if wanted.
  • Browser AI without terminal access: GitHub Use this template into a private or otherwise controlled repository.
  • Coding agent or terminal-capable AI: the agent can normally perform the clone/setup directly, subject to operator approval and repository permissions.
  • Local-only use: clone locally; no private remote is required.

For exact beginner-friendly steps and example terminal commands, see docs/ACQUISITION.md.

The acquisition method does not change how LabNote works after the copy exists.

2. Bootstrap The Workspace

  1. Open AI_ENTRYPOINT.md with the AI that will use the workspace.
  2. Confirm that the expected LabNote structure is present.
  3. Confirm whether the current workspace is private/controlled for live work or public/reference-only.
  4. Follow the lobby reading order: lobby/README_FIRST.md -> lobby/VISITOR_CHECKLIST.md.
  5. Confirm the current-run visitor handle supplied by the human operator. If none is supplied, stop and ask before writing.

No repository rename or LabNote identity-file edit should be required before first use.

3. Operate LabNote

Follow the routine deposit cycle for each contribution below. Only one writer may deposit into a workspace at a time: the human authorizes the turn, waits for destination verification and the completion report, then authorizes the next. A visitor handle identifies a session; it does not reserve a turn. Reading and preparation may happen in parallel.

  1. Copy templates/datadrop_packet.md for a new packet.
  2. Fill in the packet header and task sections.
  3. If the packet depends on larger files, add an operator-approved reference and a short summary instead of committing the raw dump.
  4. Create a JSON packet record under registry/packets/<year>/.
  5. Give the packet to the target assistant session.
  6. Copy templates/ai_response_packet.md for the answer.
  7. Create a JSON response record under registry/responses/<year>/.
  8. Review the response before marking anything accepted.

Small files, clear labels, no mystery memory. That is the trick.