CLI Reference
All 14 commands to control your brain.
brainbase init
Interactive setup wizard. Detects installed AI providers, configures integrations, sets up your LLM for knowledge extraction, and creates the brain database.
$ brainbase init
brainbase status
Shows brain health at a glance — initialization state, database status, watcher state, connected providers, and core statistics (nodes, edges, patterns).
$ brainbase status
brainbase search
Search the knowledge graph using spreading activation. Finds nodes matching your query and triggers connected memories — the same way thinking about "coffee" activates "morning" and "productivity".
$ brainbase search "typescript generics"
Options
brainbase list
Shows recent memory nodes with their importance level, type, content snippet, creation date, and confidence score.
$ brainbase list
$ brainbase list --type preference --limit 10
Options
brainbase hot
Shows the most activated nodes right now and generates a context briefing — the same narrative your AI receives. Useful for understanding what the brain is "thinking about".
$ brainbase hot
$ brainbase hot --detail MAXIMUM --topic "auth system"
Options
brainbase stats
Detailed brain statistics — total nodes, edges, active nodes, patterns, sessions. Includes type breakdown with bar charts, edge type distribution, latest session info, and critical period indicator (shows if your brain has fewer than 20 sessions).
$ brainbase stats
brainbase add
Manually add knowledge to your brain. Useful for seeding important facts, preferences, or project context. Automatically links to existing related nodes.
$ brainbase add "I prefer functional components over class components"
$ brainbase add "Project uses PostgreSQL" --type fact --importance 0.9
Options
brainbase forget
Delete specific memories from the knowledge graph. Searches for nodes matching the query and removes them permanently. GDPR-compliant — your right to be forgotten.
$ brainbase forget "old project name"
brainbase watcher
Control the background daemon that processes memories, runs consolidation cycles, and keeps the brain alive. Runs on HTTP port 7899.
$ brainbase watcher start # Start the daemon
$ brainbase watcher stop # Stop the daemon
$ brainbase watcher status # Check status + LLM connection
brainbase consolidate
Trigger a manual consolidation cycle. Normally runs automatically every 6 hours. Prunes weak edges, promotes important nodes, merges duplicates, finds creative connections (REM-style), detects knowledge gaps, and distills clusters into schemas.
$ brainbase consolidate
Edges pruned: 12
Nodes promoted: 3
Nodes merged: 2
Patterns created: 1
Dream edges: 4
Duration: 2.3s
brainbase verify
Run a 6-step brain integrity check. Validates config, database (15 tables), core nodes, hooks, signal strength, and spreading activation. Useful for debugging or after updates.
$ brainbase verify
[1/6] Config .............. OK
[2/6] Database (15 tables) OK
[3/6] Core Nodes .......... OK
[4/6] Hooks (4 events) .... OK
[5/6] Signal Strength ..... OK
[6/6] Spreading Activation OK
brainbase snapshot
View the brain's growth over time. Shows a timeline of snapshots with node count, edge count, patterns, and top topics at each point.
$ brainbase snapshot
$ brainbase snapshot --count 20
Options
brainbase dashboard
Launch the interactive 3D brain visualization in your browser. Shows the knowledge graph as a force-directed network with real-time activation levels, brain health metrics, and session history.
$ brainbase dashboard
$ brainbase dashboard --port 8080
Options
Exposes 7 API endpoints: /api/graph, /api/expertise, /api/sessions, /api/patterns, /api/snapshots, /api/activity, /api/health
brainbase embed
Manage vector embeddings for hybrid search (spreading activation + semantic similarity). Uses text-embedding-3-small.
$ brainbase embed status # Show embedding coverage %
$ brainbase embed backfill # Create embeddings for unembedded nodes
$ brainbase embed backfill --batch 50