What does this skill do, and when should you use it?
Short Video Scripter is the Social Craft skill in Aaron Marketing Skills for TikTok, Instagram Reels, YouTube Shorts, 抖音, and 微信视频号. It structures scripts around four retention gates: a 0–2 second hook, 2–5 second confirmation, 5–15 second payoff, and a loop or CTA close. Each package includes scene lines, muted-viewing on-screen text, asset keywords, platform format parameters, and an AI-content disclosure when applicable. It produces a script specification only; rendering, voice generation, editing, publishing, and scheduling remain outside its scope.
It reads a user-provided idea, draft, or transcript together with a read-only voice-card pointer, active-handle facts, platform norm cards, approved claim wording, and user-provided retention exports. For each target platform it produces timestamped rows for the four retention gates, with a spoken line, on-screen text, and shot or asset keywords; adds 2–3 hook options from named hook families; and supplies caption, first-comment, subtitle, safe-zone, and 9:16 parameter notes. TikTok and YouTube Shorts parameters cite dated norm cards, while Reels, 抖音, and 视频号 parameters are labeled Estimated. Realistic synthetic media receives an AI-content disclosure line by default, and unsupported product or offer claims are marked [needs source].
- A short-form creator needs a product-demo idea converted into a shootable 抖音 or 微信视频号 storyboard.
- A social team sees viewers dropping off in the opening seconds and wants revised TikTok, Reels, or Shorts hook and confirmation beats.
- A brand team wants one concept adapted across TikTok, Reels, and Shorts with platform-specific 9:16 parameters and copy placement notes.
- A team using an AI avatar, cloned voice, or photorealistic generation needs the disclosure line built into the video script.
How do you install this skill?
- Do not treat missing TikTok, Reels, 抖音, or 视频号 norm cards as verified current specifications; retain the Estimated label where a dated card is absent.
- Before enabling host memory or registry writes, verify host permissions, the authorized proposal flow, and recovery procedures; this skill does not itself demonstrate that the shared mechanisms are operational.
- SECURITY.md lists 17.x as the supported line while the skill and README declare 18.0.0; confirm the actual release and maintenance status.
- AI disclosure, product-claim, and platform-compliance decisions still require human or specialized review and should not rely solely on this script-writing skill.
- Shell / CLI
- Local filesystem
Python standard library scripts
Install the repository as a collection. In Claude Code, run /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then /plugin install aaron-marketing@aaron. On other Agent Skills-compatible hosts, run npx skills add aaron-he-zhu/aaron-marketing-skills. You can also clone https://github.com/aaron-he-zhu/aaron-marketing-skills. The repository is licensed under Apache-2.0.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/social/craft/short-video-scripter" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Provide the video idea or draft, target platforms, goal (follow, save, click, or watch-full), and the channel voice-card pointer. Example: Script a 30s version of this idea for TikTok and Reels: [idea]. Goal: saves. Use our voice card. The skill also documents a Chinese prompt for adapting a product demo to 抖音 and 视频号 with timestamped shots, on-screen text, asset keywords, and an AI disclosure. Without an idea or draft it returns NEEDS_INPUT. The finished package can be handed to social-quality-auditor for the pre-publish gate.
What are this skill's strengths and limitations?
- Uses a concrete four-gate retention structure covering hook, confirmation, payoff, and loop or CTA.
- Delivers scene lines, muted-viewing text, asset keywords, and platform adaptation notes in one package.
- Defaults to AI-content disclosure for realistic synthetic media and flags unsupported product claims for sourcing.
- Clearly limits itself to a script specification rather than automating production or publishing.
- Only TikTok and YouTube Shorts have dated norm cards; Reels, 抖音, and 视频号 parameters remain Estimated with named official sources.
- Requires a user idea or draft, target platforms, goal, and voice-card context; a missing target-handle dossier can produce NEEDS_INPUT.
- It does not provide compliant real-time analytics reads for closed platforms; those numbers must come from user exports.
- It does not shoot, render, edit, generate TTS, publish, or schedule videos.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Short Video Script Studio this page | 61 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Wonda CLI — AI Content Creation from the Terminal | 37 · Not recommended | ★ 158 | 14d ago | — |
| Platform Norm Profiler | 60 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Evergreen Social Calendar | 59 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Channel Portfolio Planner | 57 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
This skill is specifically for short vertical-video scripts. Creator-facing video briefs belong to brief-generator; long-form video SEO belongs to content-writer; non-video social packages belong to social-creative-builder; and calendar slotting belongs to social-calendar-builder.
How did FollowSkills review this skill?
The skill treats pasted user material as untrusted, limits itself to script specifications, performs no rendering, TTS, publishing, or upload automation, asks for confirmation before saving, uses proposal-only registry writes, and marks unsupported product claims as [needs source]. Seven points are deducted because memory and platform-data boundaries depend on the host and omitted shared protocols, rollback is not fully specified, publisher identity is unverified, and the cited regulations and platform rules cannot be verified from the supplied files alone.
The contract defines input confirmation, NEEDS_INPUT behavior, four retention beats, Estimated labeling, termination rules, and prompt-injection handling. Eleven points are deducted because this is a static review with no key-path reproduction, referenced files and runtimes are absent, and SECURITY.md calls 17.x the current supported line while the skill declares version 18.0.0, creating a version-consistency concern.
Triggers for short-video scripting, early-retention repair, and cross-platform vertical adaptation are clear; Chinese platforms are included, and non-fit cases cover creator briefs and long-form video SEO. Three points are deducted because several platforms lack dated norm cards and require Estimated parameters, while host, audience, and non-Chinese boundary evidence is limited; geo-relevance is also declared low.
The file provides structured frontmatter, Quick Start examples, a Skill Contract, data-source rules, numbered instructions, next-skill routing, versioning, and Apache-2.0 licensing, with explicit input/output and registry constraints. Four points are deducted because maintainer ownership, a concrete update path, changelog evidence, and compatibility policy are not sufficiently shown; the documentation relies on omitted cross-references and conflicts with the stated supported-version line.
The expected package includes timestamped beats, spoken lines, muted-viewing text, asset keywords, platform parameters, hook options, CTA notes, and disclosure handling, making the format directly usable for a user’s shooting and editing workflow. The static-calibration ceiling limits this to 7; eight points are deducted because no representative output or third-party execution evidence is supplied, comparative benefit over manual work is unverified, and the workflow depends on user-provided voice, claims, norm, and analytics inputs.
The skill distinguishes dated, Measured, Estimated, and [needs source] material and names supporting references; the repository also supplies general tests and CI artifacts. Six points are deducted because the supplied tests do not cover this skill’s key paths, the referenced materials are not included, and no third-party execution evidence permits independent reproduction.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →