brain-mcp

Evaluate a decision against the user's stated principles + related past thinking. Loads principles from YAML, matches by keyword, surfaces semantically similar prior decisions.

Raw accounting of what's in the brain. Counts messages, embeddings, summaries, domains. Use when user asks "how much is in my brain" or "what's the size of the corpus."

Surface patterns in how the user thinks — cognitive pattern, problem-solving approach, emotional tone, content category. Optional domain scope. Use when user asks "when do I think best" or "what patterns show up in breakthroughs."

Full re-entry brief for a domain — recent summaries + accumulated open questions, decisions, insights, quotes. Longer and deeper than tunnel-state. Use after days/weeks away from a project.

Browse conversations by calendar date. Use when user asks "what was I working on last Thursday" or specifies a date.

Surface abandoned domains with unresolved breakthrough/significant open questions. Ranked by breakthrough count → question count → conversation count. Use when user feels important threads have been dropped.

Fetch the full message thread of a specific conversation by ID. Returns role + content + timestamp, truncated at 20 messages / 1000 chars per message. Use when the user references a specific conversation from prior search results.

Return the full body of one principle by name. Use when user references a specific principle (e.g., "SHELET", "Bottleneck Amplification").

Query the user's GitHub repo + commit index. Modes timeline, conversations, code, validate. Use when user references a repo, commit, or wants to cross-reference code with conversations.

List the user's stated principles from the configured YAML file. Use when user asks "what are my principles" or before alignment-check.

Global unfinished-work matrix — every domain × every open question. Paired with dormant-contexts for the complete overwhelm-reducer surface.

Query optional analytics parquets — timeline, stacks, problems, spend, summary. Use only when user asks about tool-stack adoption, spend, or temporal patterns.

Deterministic keyword retrieval (ILIKE) over raw L0 messages. Use when the user references an exact term, name, or phrase they remember writing.

Retrieval over the user's markdown corpus (notes, wikis, docs) with filters for depth-score, breakthroughs, project, and open TODOs. Use when user references a markdown file or wiki concept.

Hybrid retrieval (vector + FTS) over L2 structured summaries. Supports filter by domain, importance, thinking-stage, source, and extract mode. Use when user wants structured knowledge, not raw messages.

Deterministic vector retrieval over embedded messages. Finds conversations by meaning, not keyword. Use when the user's phrasing may not match what they wrote at the time.

Quantify cognitive cost of switching from current domain to target domain. Formula-driven (0.0-1.0). Returns recommendation. Use BEFORE committing to a domain switch.

Show how the user's thinking on a topic evolved over time. Views full (timeline + stages), velocity (accelerating/stable/declining), first (genesis moment). Use when the user might have already moved past where they seem to be now.

Proof the safety net is intact. Structural integrity check across L0-L2. Use when user is anxious about what was lost or wants to verify nothing slipped.

Show aggregated history of a domain — thinking stages, importance breakdown, cognitive patterns, emotional tones, top concepts as bar charts. Use when user wants the meta-view of their own engagement with a topic over time.

Reconstruct cognitive save-state for a domain. Returns thinking-stage, open-questions, decisions, concepts, emotional-tone — every claim carries a citation. Use when user returns to a topic.

Conversations still in exploring or crystallizing stage with open questions at or above min_importance. Use when user feels there's unfinished work but doesn't know where.

Cross-source retrieval combining conversation embeddings, GitHub commits, and markdown docs into one ranked list. Use for broad orientation queries.

Synthesize the user's views on a topic across 10+ conversations. Modes synthesize (default) or precedent. Use when user is forming an opinion and may have already concluded something.

Monthly snapshot — activity level, top conversations, sample questions. Use when user wants to reconstruct a specific month's mental state.