Pick the Right Short-Video Hook Type: ra-hook

Where this fits

A creator drops a script in your agent and says, '这个开头太平了，帮我重写' or simply '帮我起个钩子'. Most agents default to generic question hooks or clickbait openings. ra-hook fixes that with a decision flow: identify the video's strongest asset, pick one of 7 hook types by scenario, and draft the first 1-3 sentences from that type's template. It chooses the type; a companion skill, dbs-hook, polishes execution afterward.

Why agents benefit

• Scenario-first taxonomy: 7 hook types grouped by what the hook hangs on — content, audience state, or a concrete event — so the agent matches a situation instead of guessing.
• Decision flow with fallbacks: The 'what's the hardest thing in this video?' check plus per-type prerequisites lets the agent discard unsuitable types and switch, not force a template.
• Copy-paste sentence templates: Each type ships with sentence templates and a first-3-sentences format, so agent output is structurally correct, not abstract advice.
• Anti-clickbait guardrails: Explicit rules — a promise must be kept, hype without payoff is deception — prevent the agent from generating misleading hooks.
• Clean tool boundary: The README states ra-hook decides type and dbs-hook polishes drafts, so an agent's tool-selection logic stays unambiguous.

Practical scenarios

Rewriting a script with a fresh opening (ra-洗稿)

When the agent rewrites an existing script, the new opening must avoid the source's phrasing. ra-hook re-derives the hook type from the rewritten content's core asset and generates a different but fitting opening.

Topic deepening (选题深化)

The user offers a broad subject like '为什么周一请假的打工人最多'. The agent asks what's strongest — audience pain, a shared phenomenon, or an event — then drafts a hook that puts that asset first.

First 5 seconds for a talking-head video

A user panics: '视频开头怎么做?' The agent classifies the scenario (e.g., audience pain-point for a productivity video), applies the 直击痛点 template, and produces a direct-address opening with a keepable promise.

Add it to your agent workflow

Install with:

npx skills add Pluviobyte/rnskill --skill ra-hook

Usage example: the user says '帮我起个钩子，讲新手做跨境电商踩的坑'. The agent runs the decision flow — the content is not rare, but the audience clearly has pain — selects 直击痛点, and drafts 1-3 sentences from its template, checking prerequisites before finalizing.

Compare before adopting

Compare ra-hook against generic hook-formula prompts. Three checks matter: quality signals — 1,421 stars and a paired dbs-hook skill indicate real adoption; maintenance freshness — short-video platform behavior shifts, so check the repo's recent commits for taxonomy updates; workflow fit — if your task only needs critique of an existing draft, start with dbs-hook. You can review the full skill card, including README details and templates, at https://www.openagentskill.com/skills/pluviobyte-ra-hook before wiring it in.

Why it is worth tracking

ra-hook is a well-scoped domain skill: it refuses to be a Swiss-army knife and explicitly hands polishing to dbs-hook. The star count signals traction among Chinese short-video content tooling, and the scenario-based taxonomy turns vague 'make it more engaging' requests into a deterministic agent workflow. Track it if you build agents for Chinese social platforms or video scripting; reevaluate when the repo adds new hook types or revises its decision flow — that's the signal the author keeps pace with platform trends.