prompt-preflight-qa

Prompt Preflight QA

Use When

• A storyboard prompt or full video request already exists.
• You need to catch weak first frames, missing constraints, drift risk, bad product timing, or generic ad-like language before spending credits.
• The output should be a concise readiness report, not creative approval.

Do Not Use When

• Do not use this after generation to judge finished creative quality.
• Do not rewrite the whole prompt unless the user asks; identify blockers and hand off to the right upstream skill.

Core Rule

Judge prompt quality by controllability, not by impressive prose.

Default Workflow

1. Classify the prompt object: storyboard grid prompt, full video request, no-reference draft, or reference-led replication draft.
2. Check opening strength, first-three-second hook legibility, viewer question, visible evidence, promise delivery, product timing, negative constraints, reference contract, explicit reference bindings, UGC realism, output format, and long-video segment plan.
3. Create a machine-readable preflight report when it is useful.
4. Return a concise verdict with risks, missing fields, likely drift, fixes, and whether the prompt can run now.

Blame Rule

Point the weakness to the correct upstream stage:

• bad route -> media-router or video-request-architect
• missing or misfit hook logic -> video-request-architect
• weak hook decode -> reference-decode
• weak panel, beat, or request logic -> video-request-architect
• unclear learn/do-not-copy boundaries -> reference-contract-builder

Output Shape

The artifact contains verdict, canRunNow, missingFields, majorRisks, likelyDrift, and fixNow.

Stop Conditions

• Stop when required user intent, source evidence, or owned input artifacts are missing and guessing would change the result.
• If an owned CLI or script command fails, report the exact error and stop. Do not bypass the failure with metadata-only answers, readiness probing, local payload rewrites, fallback providers, or unpublished tools.

Handoff

• Return report.json or the stdout JSON plus the smallest upstream fix.
• After fixes, rerun this preflight before generation.

Public Command Boundary

• Choose the smallest matching command or workflow from the user input and run it directly.
• If an owned CLI or script command fails, report the exact error and stop. Do not bypass the failure with metadata-only answers, readiness probing, local payload rewrites, fallback providers, or unpublished tools.
• This public skill is instruction-driven. Produce the artifact described by the workflow directly from the available evidence.
• Do not call private provider/runtime paths or unpublished local tools.
• If the CLI returns a quote-confirmation challenge, run postplus quote confirm --json --challenge-file <challenge.json> and retry with the returned token.