skills/pixverse-video-script/references/genre-routing.md

# Route A Complete YouTube Script

First identify subject, viewer promise, narrative engine, evidence spine and host stance.
Choose ONE genre by outcome, not nouns: an AI/company/money subject can use any appropriate
engine. Name the closest alternative and why it loses in working notes. No genre is a
fallback for an unclear brief; resolve a material outcome once, or infer when delegated.

| Viewer promise and evidence | Read |
| --- | --- |
| What should I do to improve/earn/build/decide? Advice, framework, operator evidence | `./business.md` |
| Which product/tool should I use; what does this actual test/demo show? | `./tech.md` |
| What is wrong/brilliant about this concrete artifact? Explicit stance/close reading | `./commentary.md` |
| What happened, how did it rise/fall, why did it matter? Verified chronology | `./documentary.md` |
| How was this specific crime/scam/legal case solved? Case records | `./crime.md` |
| How/why does this mechanism work? Causal explanation, demonstration or analogy | `./explainer.md` |

Read the selected guide fully. Do not blend six beat maps. A host/production workflow can
change headings or cue syntax while preserving that genre's hook/evidence/retention/ending.
Hook-only, advertising and fictional scene writing keep their narrower workflow; do not
force this complete YouTube contract on them.

Never invent purchases, tests, spending, credentials, revenue, channel history, confessions,
private thoughts, quotes or factual scenes. Use user receipts, named sourced evidence or
neutral analyst framing. Treat examples as craft, not facts. Mark uncertainty and verify
factual claims through supplied/primary sources. No unrequested sponsor/affiliate/course/
merch integrations. Any requested integration stays outside the chosen climax/test/payoff.

Direct output: three distinct hook options, ONE complete timestamped spoken script with
actionable production cues, and a timestamped retention map. Choose a strongest hook for
the full script when selection is delegated; do not stop after offering three openings.
Use roughly 140–150 English words/min unless the genre calls for another pace; localize
by spoken duration. Timestamps are estimates until recorded. Keep spoken copy, cue and
source notes separate. Track actual planted loops and payoffs; never write a false callback.
Do not add self-grading claims such as “high retention” in place of actual craft.