| name | wisdom |
| description | Use when prior notes, durable project knowledge, or vault-backed context should influence the current task. |
| argument-hint | [knowledge question, doc task, or note synthesis ask] |
Agmo Wisdom
Use this when durable knowledge should shape the current run.
Agmo treats the vault-backed llm-wiki as the startup-context SSOT. Session start injects a compact manifest by default; use agmo wiki context --project <project> --full for lazy full-body loading, agmo wiki maintain --project <project> for health checks, and agmo wiki migrate-wisdom --scope <user|project> --project <project> --dry-run before migrating legacy .agmo/memory/wisdom.json entries.
Main-session contract
The main session should delegate this workflow to agmo-wisdom, then integrate the result back into the active workflow.
After the delegated result is integrated, call close_agent for the agmo-wisdom lane so the completed native subagent session releases its thread slot.
Use it for:
- doc or knowledge synthesis
- note consolidation
- research summaries grounded in durable project context
- save-ready note preparation when persistence is likely
Recommended output modes
Prefer one of these explicit shapes instead of a loose freeform answer:
retrieve — find the best prior note candidates
synthesize — summarize or compare durable context
prepare-save — shape the result so it can become a durable note
Durable-knowledge contract
A good wisdom result should usually include:
- a short answer first
- evidence vs inference when interpretation is involved
- cited note candidates or knowledge sources with why they matter
- inferred project / design / plan lineage when relevant
- a recommended durable title if the result should be saved
- recommended aliases, tags, parent, and related links when persistence is likely
Artifact save body
Before ending a meaningful wisdom stage, make the final response or delegated result save-ready:
- knowledge question or durable decision
- evidence vs inference
- canonical note target and duplicate handling
- recommended note update or follow-up handoff
Canonical-note behavior
If multiple note candidates overlap heavily:
- identify the strongest canonical note
- explain why the others are weaker, duplicate-like, or draft-like
- recommend extending the canonical note instead of creating another near-duplicate unless the artifact is materially distinct