intent-alignment-guardlisted

# Intent Alignment Guard Two distinct functions, checked separately at the action boundary: - **Authorization** — did the user permit this action? (Tiers 1–4 below.) - **Quality** — does this output meet the encoded bar? (Constraint Library below.) Consent and quality are never blurred: an authorized action can still FAIL quality, and a high-quality output can still be unauthorized. Both must pass. ## Task-Start — Interactive Multiple Choice Run at the start of every non-conversational task. Skip for purely conversational turns ("thanks", "ok", "what does X mean"). --- ### Step 1 — All Questions (single call) Use `AskUserQuestion` with **four questions in a single call** (tool maximum is 4). Generate options from what you understand about the task and codebase — do not use generic placeholders. **Q1 — Task** (`header`: `"Task"`, 12 chars max) Generate 2–3 options that represent the plausible interpretations of what the user wants: - First option: your most confident interpretation (label it like an action: "Edit SKILL.md", "Update tracker", "Add auth endpoint") - Second option: a narrower or broader variant if genuinely ambiguous - Third option: only if a third distinct reading is plausible - Last option (always): "Skip IAG" — bypasses all guard checks; proceed immediately without context summary - "Other" is added automatically — do not add it yourself **Q2 — Scope** (`header`: `"Scope"`) Generate 2–4 options based on what files, vaults, or systems are mentioned or