/brainiac

/brainiac — Query and Manage the Knowledge Graph

You interact with the Brainiac knowledge graph engine. The graph stores all cross-project learnings as interconnected nodes with semantic, temporal, causal, and entity edges.

System Location

• Engine: ~/.claude/knowledge/brainiac/
• Graph data: ~/.claude/knowledge/graph/ (nodes.json, edges.json, embeddings.npz)
• Views: ~/.claude/knowledge/views/ (auto-generated markdown)

Command

Parse the user's message after /brainiac and run the appropriate command.

Important: Only accept these known subcommands: search, stats, add, link, consolidate, render, quality. Reject any other input to prevent command injection.

cd ~/.claude/knowledge && python -m brainiac <subcommand> [args]

Available Operations

If no arguments provided, show this help:

• search "QUERY" — Semantic search across all nodes
• stats — Graph overview (node counts, edge counts, quality)
• add <type> "content" — Add node (types: pattern, antipattern, workflow, hypothesis, solution, decision)
• link <id1> <id2> <relation> — Link nodes (relations: semantic, temporal, causal, entity)
• consolidate — Find merge/prune candidates
• render — Regenerate markdown views
• quality — Get quality score (0-100)

When to Use

• Before starting work: search for relevant patterns or solutions
• During debugging: search for known solutions
• When making architecture decisions: search for prior decisions
• When reviewing: check antipatterns
• Periodically: run consolidate to keep the graph healthy