entity-optimizer

# Entity Optimizer > ⚠️ **PARKED — NOT WIRED (2026-07-17 wiring audit).** No skill, orchestrator stage, or > hook routes to this subskill; `memory/entities/` has no writer today. This document is > design, not behavior (Rule 6). Wiring the Wiki-Phase entity feature (or deleting this > pair) is a product decision — see CHANGELOG [3.40.0]. Sole writer of `memory/entities/{id}.md`. ## 47 entity signals (per seo-geo) Each entity entry tracks: - name + aliases - type (Organization / Person / Place / Product / Concept) - @id (canonical URL or Wikidata QID) - description (definitive) - founding date (if Org) - industry - key relationships (sameAs, subsidiaries, etc.) - ai_resolution_status per engine (chatgpt/pplx/claude/gemini) - last_probed_at - citation_count_30d - (... 47 total) ## Workflow ``` 1. After research / draft / publish, scan for entities mentioned 2. For each new entity: - Create memory/entities/{slug}.md - Populate from Wikidata if available - Probe AI engines for recognition 3. For each existing entity: - Update citation_count - Re-probe if ai_resolution_status known stale (>30d) 4. Cross-project linking: entities are global (used across all projects) ``` ## ai_resolution_status states - recognized: brand named accurately + sources cite our URL - partial: brand named but facts wrong OR our URL not cited - unrecognized: brand not mentioned - confused: brand mentioned but mixed with another (HIGH RISK) ## See also - `references/geo/47-entity-sig...