brain-bootstrap

name

brain-bootstrap

description

Set up this machine and this agent to participate in the Obsidian-Vault brain — clone vault if needed, register the machine, install agent-flavor configuration, write Current Activity, verify. Use when the user says "bootstrap this machine", "set up the brain on this machine", "onboard this agent", "this is a new machine — set it up", "install the brain on this computer", "make this machine part of the brain", or any first-time brain installation request. ONE-SHOT setup, autonomous after the human answers vault path and machine name.

brain-bootstrap

One-shot bootstrap for getting a new machine + agent fully participating in the Obsidian-Vault brain.

When to invoke

• User says any of: "bootstrap this machine", "set up the brain", "set up the brain on this machine", "onboard this agent", "this is a new machine — set it up", "install the brain on this computer", "make this machine part of the brain"
• A new Claude Code session on a machine that has never been registered in AI Brain/Machines/
• A human is bringing up a fresh laptop/desktop and wants it talking to the same vault as their other machines

What to do

The full runbook is in the vault at ~/Obsidian-Vault/AGENT-BOOTSTRAP.md. Read it once, execute it top-to-bottom.

Quick orientation:

1. Read the runbook first: Read ~/Obsidian-Vault/AGENT-BOOTSTRAP.md (or the equivalent path on this machine — if vault isn't cloned yet, you'll have to read this from somewhere else first; in that case, ask the human to paste the runbook content or point you at the GitHub URL: https://github.com/YOUR-USERNAME/obsidian-cortex/blob/main/AGENT-BOOTSTRAP.md)
2. Run the four self-check detections at the top of the runbook
3. Decide which branches apply using the decision table
4. Pause and ask the human exactly once for: vault path (if not at default), machine name. After those answers, proceed autonomously through the chosen branches.
5. Verify with status and snapshot at the end
6. Commit and push the machine setup
7. Hand off to AGENTS.md for ongoing protocol

Rules

• ONE pause-and-ask for the human at the start (vault path + machine name combined into a single question if possible). After that, execute the rest autonomously.
• Never invent paths or machine names. Use what the human says verbatim.
• Never blanket git add . — only add AI Brain/Machines/<MACHINE>/ and any other paths you explicitly created during bootstrap.
• Stop and ask if any verification step fails — do not improvise around errors.
• If snapshot shows another machine ACTIVE on the same project, tell the human before continuing.

When to NOT invoke

• If AI Brain/Machines/<thismachine>/ already exists and snapshot shows recent activity — this machine is already bootstrapped. Invoke brain-startup instead to begin a normal session.
• If the user just wants to start a session on an already-configured machine — that's brain-startup, not this skill.

This skill is for first-time machine setup only. Once bootstrapped, the normal session lifecycle is brain-startup → work → brain-closeout → (optional) brain-daily.

After successful bootstrap

Brief the user:

"Bootstrapped <MACHINE> into the brain. Vault is at <VAULT>. Snapshot shows N machines registered: . Ready to work — what would you like to do?"

Then the agent is in the same state as if they'd just run brain-startup on an existing machine. The user can immediately give a real task.