clarify

# Clarify ## Consequential advice For a consequential recommendation, follow the `Advice gate` in `dissenter`. It overrides the local permission to recommend an answer when the user is stuck. When the gate applies, first say that you are using `/dissenter` and why. Help turn a vague idea into clear requirements through natural collaborative dialogue. Start by understanding the current project context, then ask questions one at a time to refine the idea. Keep interviewing until purpose, constraints, and success criteria are thoroughly clear — then output a copy-paste prompt, append the configured completion suggestions, and stop. <HARD-GATE> Do NOT write code, scaffold projects, propose implementation plans, write design docs, commit specs, or invoke writing-plans. This skill ends with a single copy-pasteable prompt plus its configured completion suggestions. </HARD-GATE> ## Checklist Complete these in order: 1. **Explore project context** — check files, docs, recent commits 2. **Scope check** — if the request spans multiple independent subsystems, flag it and interview one slice at a time 3. **Ask clarifying questions — extensively, one at a time** — purpose, constraints, success criteria, non-goals, priorities, edge cases, users/audience, what "done" looks like 4. **Stop when thorough** — all three pillars are specific enough to act on, not just stated once 5. **Deliver copy-paste prompt** — before drafting an applicable coding prompt, load `${AGENTIC_HOME:-$HOME/.agentic}/skills/pattern-gate/SKILL.md` and include its pattern list and approval requirement inside the prompt. Deliver one self-contained fenced prompt, followed by the `clarify` completion suggestions from [skill connections](../../../docs/skill-connections.md); stop ## Question discipline - **One question per message** — if a topic needs more exploration, break it into multiple questions - **Multiple choice preferred** when it speeds answers; open-ended is fine when needed - **Explore the codebase** when a question can be answered from the repo instead of asking the user - **Go deep on branches** — constraints often hide in edge cases and non-goals **Extensive bar:** each pillar should survive "what about when…?" Shallow one-liners mean keep asking. ## Scope check Before detailed questions, assess scope. If the request describes multiple independent subsystems (e.g., "build a platform with chat, file storage, billing, and analytics"), flag it immediately. Don't spend questions refining details of a project that needs to be decomposed first. Help the user pick one slice to clarify now. Other slices get their own clarify session later. ## Drawing the user out Unlike `grill-me`, do not lead with recommended answers. Draw the user's thinking out with questions. If they are stuck on a specific decision, a short recommendation is fine — then return to asking. ## End artifact When purpose, constraints, and success criteria are thoroughly clear, your **final message contains only a fenced code block** with a self-contained prompt the user can copy-paste into a new session, followed by the `clarify` completion suggestion block from [skill connections](../../../docs/skill-connections.md). No preamble, no recap, and no other text outside the prompt and suggestion block. Weave into the prompt: what to build, why, constraints, success criteria, non-goals, and any unresolved open questions. Write it as instructions to a fresh agent — imperative, specific, complete enough to act on without this conversation's history. For applicable coding work, include a **Pattern list** section using `pattern-gate`'s name, where, and why format. The list belongs inside this prompt, not in a later implementation session. Include an instruction to obtain approval of the prompt and list together before editing. List unresolved choices as open questions; this remains a requirements prompt, not an implementation plan. Example shape (adapt to the actual topic): ``` Build [what]. The goal is [purpose]. Constraints: - [constraint] - [constraint] Done when: - [success criterion] - [success criterion] Out of scope: - [non-goal] Open questions (resolve if ambiguous): - [question, if any] ``` No file write. No git commit. ## Key principles - **YAGNI in questioning** — ask about what matters for the current slice, not every hypothetical feature - **Incremental clarity** — each answer should sharpen the picture; vague answers get follow-ups - **Be flexible** — go back and re-ask when something new contradicts an earlier answer