prompt-enrichmentlisted

# Prompt Enrichment — Structured Prompt Builder Transform vague user prompts into structured, actionable prompts that reduce miscommunication and produce better outcomes. Uses a 4-part framework with technique selection. ## When This Activates Called by the `prompt-enrich-trigger.js` UserPromptSubmit hook, which injects a `[PROMPT-ENRICHMENT-GATE]` instruction into Claude's context whenever a user prompt is classified as vague. ## Step 0: Look up existing patterns FIRST Before building a new enrichment, check if a similar pattern is already stored. Run: ```bash node ~/.claude/packages/kernel/spawn-state/prompt-pattern-store.js lookup --raw "<raw user prompt>" ``` This returns JSON. Behavior depends on `bestMatch` and `bestMatchTier`: - `bestMatch.score >= 0.8` AND `bestMatchTier == "Independent"` (5+ approvals) → silently apply the stored enrichment, show only a one-line summary like *"Using your established pattern for {category} (5+ approvals)."* Skip steps 1–4. - `bestMatch.score >= 0.8` AND `bestMatchTier == "Trusted"` (3–4 approvals) → show one-line summary, auto-proceed unless user objects. - `bestMatch.score >= 0.8` AND `bestMatchTier == "Familiar"` (1–2 approvals) → show stored enrichment, ask "Look right?" - No match (or score < 0.8) → continue to Step 0.5 (and Step 1) to build a new enrichment with conversation context. ## Step 0.5: Read prior conversation context (H.7.5) Before classifying or building, sample the **last 1-3 turns** of the conversation. The