Node/link counts for the whole graph, broken down by layer.

Search nodes by label or id (ranked: exact > prefix > substring). Use this first to find a node's id.

Fetch a single node (all fields + data) by its exact id.

Aggregate: filter the graph, then group and count. Use for 'how many', 'top N', 'most/least' and any date-window question — search ranks, this counts. e.g. major incidents in a date window: {type:'observation', where:{'data.kind':'incident','data.severity':'major'}, since:'2026-07-12'}; the biggest absence offender: {where:{'data.kind':'absence'}, groupBy:'data.subject'}.

The neighbourhood around a node: connected nodes + links out to depth hops.

Shortest path (chain of nodes) between two node ids; null if unreachable.

Persist a memory to the brain — one durable, atomic fact/decision worth recalling later. Written to the append-only journal beside the snapshot; immediately queryable and survives every rebuild. Author [[node-id]] links inside the text where you know a connection.

File a report — a plain-English summary of what an agent just closed. Lands on the panel's Reports timeline; durable and immediately queryable. Call this as the last act of a session.