Timps

Generate a learning plan based on what you keep asking about vs what you've decided. High gap_score = you keep asking but never decide.

Compare current code structure against past architectural decisions and detect drift.

Backfill embeddings for existing memories.

Get embedding backfill status.

Get the provenance chain for a memory — where it came from, how it was derived, and reliability scoring (Tool 20 SourceAttributor).

Export all memories as JSON.

Extract commitments from meeting notes. Use after any meeting.

Extract facts from a conversation.

Score a memory's false-memory risk based on provenance, source reliability, evidence count, and age (Tool 18).

Retrieve past architecture decisions and design choices. Use when making structural code decisions.

Retrieve personal coding patterns and conventions I always use.

Get context string for current project.

Count and analyze context switches in current session.

Get recent episodes.

Show past incidents and bugs in a specific module or area.

Get the Living Manifesto — actual values from behavioral patterns, not stated beliefs.

Get all stored memories, goals, and preferences. Use before starting a complex task.

Retrieve code patterns relevant to the current file or task context.

List all pending commitments not yet completed.

List all tracked positions with conflict counts.

Retrieve what you were working on in recent sessions. Use before starting a task to restore context.

Retrieve team-level architecture decisions. Scoped to the current project.

Get memory statistics.

Show productivity trend — am I faster or slower this week vs baseline?

Get current working memory state.

Auto-extract typed schemas from memory stream — detects repeated patterns and infers structure types (Tool 25 SchemaInferrer).

Add a behavioral observation to the manifesto corpus. Mines from positions, decisions, and architecture insights automatically on init.

List contributors who haven't been seen in 90+ days, with their contributions preserved. Use to recover knowledge after someone leaves the team.

Log a decision outcome to build the Regret Oracle knowledge base.

Log a past decision with its outcome. Seeds the simulation model used by timps_simulate_decision.

Log a question you asked. Powers CurriculumArchitect's learning-gap detection.

Look up known quirks for an API before using it.

Mark that a contributor was active on a given date. Updates their last-seen timestamp used by timps_institutional_memory.

Get the episodic graph structure (nodes + edges) for visualization. Returns recent episodic nodes and their temporal/causal links.

Ingest a signal into NexusForge episodic memory. Builds hybrid graph nodes with time-aware gists and factual links. Use after coding sessions, decisions, or notable events.

Query NexusForge episodic memory with agentic iterative retrieval. Returns episodic nodes with spatiotemporal context, refusal when evidence is insufficient.

Get NexusForge episodic memory statistics: node counts, edge counts, and source breakdown.

Add a decision text to the codebase culture corpus. Mines from positions, decisions, and architecture insights automatically.

Predict bug risk for a planned change based on past incident history.

Save a discovered API quirk for future reference.

Record a bug to build personal pattern profile.

Record a contribution from a team member. Preserves their decisions/patterns/incidents/quirks so the team can reference them after they leave.

Record an episodic event.

Record a production incident to build pattern library.

Store something you learned today for future reference.

Record that a person was mentioned in some context. Builds the relationship profile that powers drift alerts.

Store a reusable code pattern or convention for future recall.

Store something you consistently flag or look for during code reviews.

Log a behavioral signal for burnout tracking.

Record an error encountered.

Record file being worked on.

Register a prospective trigger: when a specific phrase appears in context, surface a specific memory or action (Tool 23 / L17).

Check relationship health and drift alerts.

Check for conflicts/contradictions between two memories using Jaccard similarity + sentiment analysis (Tool 21 ConflictResolver).

Analyze memory for over/under-representation bias across saved facts (Tool 24 / L18 BiasRevealer).

Search memories by query.

Set a goal in working memory.

Simulate future outcomes for a decision based on actual history.

Coach using your own observed workflow patterns. Reframes VelocityTracker advice as "how YOU do this work" rather than generic tips.

Store an important fact in TIMPs long-term memory.

Run a metabolic consolidation cycle. Consolidates high-activation nodes, audits low-utility nodes, refreshes stale nodes, and decays inactive ones.

Get the metabolic graph structure (nodes + edges) for visualization. Returns nodes with activation scores and layer information.

Ingest a signal into SynapseMetabolon spreading activation graph. Classifies into interaction/reasoning/audit layers with automatic entity linking and activation spread.

Query SynapseMetabolon with spreading activation. Spreads from seed nodes through relational edges, runs metabolic cycle, and returns activated nodes with confidence scores.

Get SynapseMetabolon statistics: node counts, edge counts, layer breakdown, and average activation.

Query memories that were valid at a specific point in time. Use for historical reconstruction ("what did I believe on March 15?"), longitudinal tracking, or context-anchored retrieval.

Verify the integrity of the immutable memory audit trail (L10 EngramLog). Returns valid/invalid and block count.

Check if coding context matches personal bug triggers. Use before writing code under pressure.