Clarify
Spec-Driven Agentic OS for Claude Code + OpenAI Codex CLI. Native skills, subagents, hooks, and an adversarial codex review loop. Inspired by github/spec-kit, built on Claude Code primitives. No daemon, no vendor lock-in.
npx -y skills add MakFly/ai-dev-kit --skill clarifyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use PROACTIVELY when the user invokes /clarify <slug> or when a spec has Open questions and the user wants to resolve them. Reads specs/<slug>/spec.md §6, asks targeted questions, integrates answers, bumps Status to `clarified`.
SKILL.md
2.2 KB, as published. Nobody here has run it
/clarify
Resolve specs/<slug>/spec.md §6 Open questions one by one with the user, then bump Status: clarified.
Procedure
- Read spec.
cat specs/<slug>/spec.md. If file missing → tell user, suggest/specify <slug>first. - Find Open questions. Section
## 6. Open questions. If empty, ask user what to clarify — do not invent. - Ask one by one. For each question, present it to the user with 2-4 concrete options + a "Other" / free-form fallback. Use
AskUserQuestiontool if available, else plain chat. - Integrate. For each answer, edit the spec:
- Move the question from §6 to the relevant §1-§7 with the resolved decision.
- Add the decision as a bullet under "Decisions resolved during /clarify" appendix (create if missing).
- Bump Status. When §6 is empty, set
Status: clarifiedand update theDate:field. - Commit.
git commit -m "spec(<slug>): clarify open questions".
Doctrine
- One question at a time. Do not batch — the user must consciously decide each.
- Concrete options. "What database?" → "(1) Postgres, (2) SQLite, (3) Bun.sql, (Other)". Never open-ended without scaffolding.
- Record reasoning. Each resolved question keeps a one-line rationale in the appendix.
- No silent choices. If the user defers, the question stays open with
Deferred: <reason>annotation.
Output appendix (added to spec.md)
## Appendix A — Decisions resolved during /clarify
- **Q (was §6):** <question>
- **A:** <decision>
- **Rationale:** <one line>
- **Date:** <ISO>
On completion
Print to chat: count of questions resolved, count deferred, new Status. Suggest /plan <slug> if Status is now clarified.