dev-planlisted

# Plan ## Success I := π written ∧ ## Task IDs section ∃ with ≥1 `T\\d+:` line V := `ls artifacts/plans/{N}-*.md*` ∧ `grep -E '^- T[0-9]+:' artifacts/plans/{N}-*.md*` Let: σ := spec artifact π := plan artifact at `artifacts/plans/{issue}-{slug}.md` τ := tier ∈ {S, F-lite, F-full} χ := open gap (e.g. leftover `[NEEDS CLARIFICATION]` in σ) Spec → micro-tasks → agent assignments → plan artifact → **Executive Summary** → free-form approve → seed + commit. **Flow: single continuous pipeline. ¬stop between steps except (a) blocking χ pre-flight prose, (b) optional `--audit` prose note, (c) Step 6 sole approval stop.** **Confidence policy:** auto when the path is unique; sole human gate at Step 6 (chat-native). ¬double-approve (no mid-plan 2f + final 6). ## Hard ban — AskUserQuestion **Never call AskUserQuestion / `present choice` / multi-select tool prompts in this skill.** Human-in-the-loop is **chat-native** (same doctrine as `/analyze` / `/spec` / `/frame`): 1. Produce the plan (auto force-splits, auto next slice when unique). 2. Print a clear **Executive Summary**. 3. **Stop this turn** and wait for the user's free-form reply. 4. Interpret natural language (approve / modify … / return to spec / adversarial / advisory) and act. No button menus. No forced option lists. Ambiguity → χ in summary or **one short prose clarifying question** then **STOP this turn** — still ¬AskUserQuestion. ``` /dev-plan --issue 42 Generate plan from spec for issue #42 /dev-