AI Engineering · Version 1.2.0 · Reviewed 2026-08-02
AI Skill Preflight Reviewer
Make AI behavior measurable and safer for skill release preflight and skill rule compliance review with evidence, explicit trade-offs, and a verification plan.
4 method steps
5 documented failure modes
5 diagnostic checks
7 quality gates
Checks a skill or scenario against structural rules, routing safety, content quality, evaluation coverage, and destination policy. It grounds the decision in the skill files, destination rules, manifests, references, scripts, evaluations, versions, and proposed change diff and explicitly prevents a format-only pass approving a skill with overlapping triggers, missing references, or untested destructive authority.
₹199 one-time
Get this skill archive
What it checks first
AI Skill Preflight Reviewer checks a skill or scenario against structural rules, routing safety, content quality, evaluation coverage, and destination policy. It grounds the decision in the skill files, destination rules, manifests, references, scripts, evaluations, versions, and proposed change diff and explicitly prevents a format-only pass approving a skill with overlapping triggers, missing references, or untested destructive authority. Use it when the work involves Skill release preflight, Skill rule compliance review, Skill promotion readiness.
- Trigger precision against neighboring skills, because an excellent method is useless when the wrong requests invoke it.
- Whether the skill encodes stable decision knowledge or merely restates a host operation that belongs in a tool or script.
- The complete authority chain from instruction to tool call, including confirmation requirements and failure propagation.
- Runtime context cost from always-loaded instructions, references, examples, and sibling overlap.
- Evaluation coverage for positive, negative, ambiguous, missing-evidence, and conflicting-skill requests.
Example task
Input
Apply the ai skill preflight reviewer to our current skill release preflight work. We need a concrete decision, bounded changes, and evidence that the result is correct.
Expected output
Start with the skill files, destination rules, manifests, references, scripts, evaluations, versions, and proposed change diff. The highest-risk failure is a format-only pass approving a skill with overlapping triggers, missing references, or untested destructive authority. Separate release blockers from warnings and require evidence for every enforced rule. Verify the result by running deterministic checks plus positive, negative, and conflict-routing cases from the final revision.