ox-consult

---

name: ox-consult description: >- Search SageOx team memory BEFORE answering from first-principles reasoning. Auto-fire when the user references recent or specific work or asks a before/after question: "I just pushed...", "did X fix Y?", "is the alert gone now?", "this request", or anything tied to a prior decision, a prod anomaly, or a metric/cost change. A confident answer that prior work contradicts is worse than a slow one — check first, then answer. Routes the cue to the right corpus: recency to ox session list, conceptual to ox query, code-provenance to ox code search, decision-record work to ox decision enrich.

<!-- Thin by design. The authoritative consult-first reflex — the cues, the "confident-wrong is worse than slow" reasoning, and the routing rationale — lives in the <consult-first> block of `ox agent prime` (Layer-1 floor), which reaches every primed AI coworker. This native skill adds discovery and activation ergonomics; it duplicates no reasoning. Every command it names is already reachable from the floor, so removing this skill loses ergonomics but never the floor. Do not grow this body. -->

Use when

The user's message carries a consult cue — STOP and search before reasoning:

• References recent or specific work: "I just pushed...", "this request", "did X fix Y?", "is the alert gone now?"
• Touches a prior decision, a prod anomaly, or a metric/cost change — anything with a before/after.

Route the cue

The authoritative cue→corpus routing (recency vs conceptual vs code-provenance, with the exact commands) lives in the <consult-first> block of your ox agent prime output — follow it from there rather than from this file. If it is not in context (session not primed), run ox agent prime first, then route.