install-cerebras-brain

name	install-cerebras-brain
description	Install or repair the runtime for the user's personal knowledge base (the 'cerebras brain') on this machine: Python venv + deps, Docker Postgres+pgvector container, database schema, Ollama + embedding model, and the `kb` MCP server registration. Use for FIRST-TIME install, after moving the project, or when start-cerebras-brain / ask-cerebras-brain report something broken that starting alone can't fix. Idempotent — safe to re-run. For the everyday 'bring it up' action use start-cerebras-brain instead.
trigger	/install-cerebras-brain

/install-cerebras-brain

Install-or-repair the cerebras brain runtime. Idempotent: every step checks-then-acts, so running it is also the way to diagnose what's missing. Day-to-day starting is /start-cerebras-brain; this is the heavier rare-use tool (it re-verifies pip deps, schema, and MCP registration).

Do this

1. Run the bootstrap from the project root:

   cd <repo>
   .\setup.ps1
   

   Steps, in order: (1) Python venv + deps, (2) Docker engine, (3) Postgres+pgvector container, (4) schema, (5) Ollama + nomic-embed-text, (6) register the kb MCP server (user scope). It ends by printing kb_health.

2. Read the output and report in plain language — which steps were already [ok], which it had to [do], and any [warn]. The final health block should show rows > 0 and warnings: none.

When it can't self-heal (report clearly, don't pretend success)

• Docker or Ollama not installed: it tries , but Docker Desktop may need a reboot / WSL2 enablement. If it warns, give the user the exact manual step, then re-run.