Problem

When an agent connects to gbrain for the first time or after an upgrade, there is no guided onboarding path. The agent wires up gbrain and never discovers critical health issues.

Real-world data from a 165K-page production brain:

Proposed: gbrain onboard

gbrain onboard          # interactive
gbrain onboard --auto   # apply all fixes
gbrain onboard --check  # dry-run

1. Runs doctor + features + health + schema stats
2. Identifies top issues by impact
3. Presents migration prompts for each fixable issue
4. Records completed migrations in kv (idempotent re-runs)
5. Suggests unused features (think, brainstorm, salience, etc.)

Migration prompt format:

id: onboard-entity-links-v1
version_introduced: 0.41.0
condition: "entity_link_coverage < 0.5"
prompt: "Entity link coverage is {coverage}%. Run extraction?"
command: "gbrain extract links --ner"
auto_eligible: true

Related PRs:

• feat: auto-link entity mentions for orphan reduction #1378 auto-link entity mentions
• feat: smart embed scheduling with priority and catch-up mode #1379 smart embed scheduling
• feat: NER-based entity link extraction #1380 NER entity linking
• feat: timeline extraction from meeting transcripts #1381 timeline extraction from meetings
• feat: takes bootstrap from existing content #1382 takes bootstrap