Tool reference — all 37

Grouped by area, as this page always has been. The server itself groups tools by moment of use instead: 26 in the menu by default (the work toolset, below with no tag) and 11 more for brain maintenance, marked (maintenance), that stay out of the menu until asked for — from the CLI (wadachi sleep, wadachi doctor) or by setting WADACHI_TOOLSETS=work,maintenance for a session. manual prints the full description of any of the 37. Args in italics are optional.

Memory

tool args does
store_memory content, title, project, tags, category save knowledge (Memories)
get_memory memory_id full content by id; counts as access
list_memories project, category browse the index
update_memory memory_id, content, tags replace — previous version kept
delete_memory memory_id permanent (prefer flag_stale)
memory_history memory_id all prior versions

Search & context

tool args does
get_context cwd, task_description, limit, max_tokens, format dense session start + proposals (get_context — start here, every session)
recall query, project, limit, neighbors semantic/keyword search, annotated (Search & retrieval)
expand_memory ids batch pointer → full content (max 10)
brain_status health, search mode, stats, projects
brain_watermark project current position (highest id per table) — take it before starting work
changed_since watermark, project what appeared after a position taken earlier
manual full description of every tool, in the menu and out

Decisions & projects

tool args does
store_decision decision, rationale, alternatives, context, project record a choice (Decisions)
list_decisions project, limit recent decisions
register_project name, description, paths enable auto-detection (Projects & scoping)
list_projects all registered

Desk

tool args does
desk action, title, objective, done_when, plan, slug, outcome, distil, cwd, project open, close, or list a desk — the plan and progress for a task in flight
desk_read slug, cwd, project pick a desk's thread back up: plan, next step, what was already tried
desk_log slug, done, note, open_question, add, cwd, project tick the step that landed, note what failed, get the next step back

Graph

tool args does
recall_associative query, project, limit spreading activation + baseline (The typed knowledge graph)
related_memories memory_id, limit strongest typed neighbours
memory_graph project, focus_id, include_entities stats, hubs, orphans, Mermaid
rebuild_entity_graph project optional entity layer (local claude CLI) (maintenance)

Beliefs

tool args does
review_beliefs project read-only stale scan (Beliefs, staleness & decay) (maintenance)
set_belief memory_id, confidence, status, valid_until, review_reason, superseded_by edit the envelope (maintenance)
flag_stale memory_id, reason, superseded_by confirm staleness — recoverable

Reflection & procedures — all maintenance

tool args does
reflect project, limit, store_them cross-memory insight candidates (Consolidation, insights & procedures)
list_insights status the proposal queue
accept_insight insight_id, project promote to linked memory
reject_insight insight_id keep as rejected
review_procedures project incident clusters → candidate rules

Consolidation & time

tool args does
sleep project, min_similarity the full housekeeping report (Sleep — consolidation, proposed) (maintenance)
consolidate project, threshold, max_groups similarity-based merge groups (maintenance)
merge_memories source_ids, title, content, project, tags your synthesis + supersession (maintenance)
why question, project, limit decision provenance (Provenance & time-travel)
as_of date, query, project, limit the brain at a past date