iterative-critiquelisted

# Skill-13: Iterative Critique & Revision Loop (The Coaching Brain) Input: `draft_text` (the Skill-7 output) + `use_case` (e.g. `Salary_Negotiation`) + `iteration_count` (1-based) + `max_iterations` (default 3) + optional `user_revision_request` Output: strict JSON — `{ "action": "display_critique", "iteration": N, "critique_points": [...exactly 3...], "question_to_user": "..." }` **or** `{ "action": "force_exit", "message": "..." }` ## Procedure 1. **Load the model's critique dimensions** — Read the `## Model reference` section in your agent instructions and extract the **Critique dimensions** (or Read `config/model-reference.md` for the same list). Done when the 3 dimension groups (model integrity · tone & audience fit · logic & persuasion gaps) are in context; if missing → stop, report `{"action": "error", "reason": "Model <agent_model> has no critique dimensions"}`. 2. **Check the iteration cap** — If `iteration_count >= max_iterations` (default 3): break the loop and return `{ "action": "force_exit", "message": "Reached 3-iteration limit. This is the best version we could generate so far." }`. Done when the cap check decides the path. 3. **Critique the draft** — Against the model's dimensions, review the `draft_text` for: (1) Model Integrity — does it follow the model's structure? (2) Tone & Audience Fit — is it right for the `use_case`? (3) Logic & Persuasion Gaps — what weakens the message? Provide **exactly 3 actionable improvement points**, each naming the specifi