decision-first

Decision-First Protocol

Makes a decision on behalf of the user via a 5-section model with full reasoning and alternatives. Invoked whenever the agent is about to ask an architectural / project / scope question — instead of asking, the skill is called.

Core idea: the user prefers autonomous decisions with transparent reasoning over back-and-forth Q&A. A decision can be contested post-hoc (round 2) — cheaper than a synchronous pause-and-ask.

When to call: exists 2+ reasonable alternatives and a justified selection is required. If one option is obviously correct — do not call, just act.

Input

An ambiguous architectural, scope, or design question the agent is about to ask the user.

Output

decision-{NN}.md capturing the choice via a 5-part model (🎯 Decision / Why / 🛡 Safety / 📈 Scalability / Alternatives / In plain language).

Hands off to

Returns control to the calling skill (typically arch-first, plan-first, or library-first).

---

Step 0 — Task context

Ensure {log_dir} and {slug} are known. Otherwise resolve from the latest task log folder.

---

Step 1 — Identify decision(s)

Auto-mode — pick the last unresolved question from the current conversation. If related questions cluster (e.g. "flatten + format" = 2 decisions) — handle all in one invocation, one artifact per decision.

Manual mode (/decision-first) — either process the questions the user described, or ask "which questions?".

Cutoff: obvious choice → skip the skill, just act.

---

Step 2 — Apply the 5-section model

Per decision:

# D{N} — {short title}

🎯 **Decision:** {chosen option in one sentence}

**Why this option:**
- {specific reason 1}
- {specific reason 2}

**🛡 Security:** {security analysis. If N/A — explicit `not applicable`}

**📈 Scalability:** {scaling impact. If N/A — explicit `not applicable`}

| Alternative | Why rejected |
|---|---|
| {option 1} | {concrete drawback} |
| {option 2} | {concrete drawback} |

**Plain-language analogy:** {real-world analogy without technical jargon}