Save a specific piece of knowledge to the Graphonomous knowledge graph.
Knowledge to store and optional flags: $ARGUMENTS
User explicitly asks to remember or save something
A decision, finding, or fact should persist beyond this session
After learning something that isn't tied to a specific retrieval (use /graphonomous:learn when it is)
act(action: "store_node", content: "<one atomic fact>", node_type: "semantic", confidence: 0.7, source: "conversation")Check for duplicates first — always:
consolidate(action: "query", operation: "similarity_search", query: "<the fact>", limit: 3)similarity > 0.90 — duplicate — update existing node via learn(action: "from_feedback", node_id: "<id>", feedback_type: "correction", correction: "<updated content>")
similarity 0.70–0.90 — related — store new node + add related edge
similarity < 0.70 — novel — safe to store
Store the node:
act(action: "store_node",
content: "<one atomic fact — keep it focused>",
node_type: "semantic",
confidence: 0.7,
source: "conversation",
metadata: {"topic": "<topic>", "session": "<context>"}
)Link to related nodes (if similarity search found neighbors):
act(action: "store_edge", source_id: "<new_id>", target_id: "<related_id>", edge_type: "related_to", weight: 0.5)| Type | Use When | Example |
|---|---|---|
semantic | Facts, definitions, architecture | "WebHost.Systems uses Supabase Auth with JWT" |
procedural | Workflows, how-to, recipes | "To deploy: run mix release, then docker build" |
episodic | Events, observations, what happened | "Auth middleware was refactored on 2026-03-15" |
temporal | Time-bound monitoring events | "CPU spike at 14:30 during load test" |
outcome | Measured results, benchmarks | "Latency dropped 40% after caching change" |
goal | Objectives, intent (prefer /graphonomous:goals for formal goals) | "Need to migrate auth to new compliance standard" |
0.9–1.0: Directly observed in code/docs/tests
0.7–0.89: Strong evidence, not directly verified
0.5–0.69: Single source, plausible
0.3–0.49: Indirect evidence, uncertain
0.0–0.29: Speculative — flag as such
For storing several related facts at once:
Store each as a separate node (one atomic fact per node)
Create edges between them:
Same topic → related_to (weight 0.5)
Cause/effect → causal (weight 0.7)
One supports another → supports (weight 0.6)
One supersedes another → supersedes (weight 0.8)
Storing without duplicate checking — always search first
Multi-fact nodes — split into atomic pieces and link with edges
Setting confidence above 0.7 for conversational knowledge — reserve 0.8+ for verified facts
Storing ephemeral session state — use tasks/plans for that, not the knowledge graph
Skipping edges — isolated nodes are hard to retrieve later