More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Video Craft: Hooks, Pacing, and Polish

    1

    The craft pass that separates a video people watch from one that merely renders: attention contract, legibility and motion rules with real numbers, sound levels, packaging, and a scored review.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Video Craft: Hooks, Pacing, and Polish

    Video Craft: Hooks, Pacing, and Polish

    Example session with this skill installed

    Review my 60-second product teaser before I publish it. It opens with our logo animation, then five product stills with captions while a music track with lyrics plays under the voiceover, which people say is hard to hear. It's vertical, for Reels.

    • Read your context and instructions
    • Compiled the video craft

    PRE-PUBLISH REVIEW — produced by the skill's review pass. Verdict up front: do not publish yet — two blockers. Finding 1 — BLOCKER, hook missing (0:00–0:05). The logo animation spends the entire hook budget before any value appears. On muted autoplay there is no text promise at all. Fix: cut the sting to a final-frame end card. Open on the strongest product shot with the promise as on-screen text — the finished result first. Finding 2 — BLOCKER, slideshow trap (0:05–0:50). Five stills with caption bars: nothing moves with intent, one transition repeats five times, and the captions are doing the visuals' job. Any single yes on the trap questions is a blocker; this draft answers yes to three. Fix: two of the five shots get regenerated or re-shot with defined first and last frames (detail → full product), shot sizes alternate wide/detail/wide, and the remaining stills get genuine camera moves, not template zooms. Finding 3 — FIX, sound (throughout). A track with lyrics under a voiceover always competes; here it's also only a few dB under the speech, which is why the voice is hard to hear on phone speakers. Fix: instrumental version of the track, ducked 18–20dB below the voice. Check overall loudness near −14 LUFS, true peak under −1dB, no clipping. Finding 4 — FIX, caption placement (0:12, 0:34). Two captions sit in the bottom ~300px where the Reels UI overlays them. Fix: all captions inside the central safe zone, at least ~48px at this frame size, held ~3 seconds per 60 characters. Finding 5 — POLISH. The teaser never drops the music, so the price reveal at 0:52 has nothing to land against. Fix: two seconds of silence under the reveal. Scorecard: hook 0, legibility 1, motion 0, sound 1, packaging 2 (aspect and first-frame intent are right once the sting moves), consistency 2 — 6/12 with zeros on hook and motion. Both zeros are fixable with the changes above; re-review the two blocker areas after the re-cut, then publish.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Plenty of videos render perfectly and still get skipped in four seconds: a logo sting where the hook should be, captions too small to read on a phone, music fighting the voiceover, and a run of stills with text bars that is a slideshow wearing a video's clothes. This skill is the craft standard that catches all of that, during production and again as a pre-publish review. It starts with the attention contract — a hook in the first three seconds, a real payoff inside thirty, one idea per 30–45 seconds, and a close that pays off the opening promise. Legibility rules come with numbers: type sizes at 1080p, the 4.5:1 contrast floor, the ~80% safe zone, and the bottom 300px of vertical video that platform UI owns. Motion discipline covers easing, staggered reveals, transitions that carry meaning, and pattern interrupts on a schedule. Sound design is equally concrete: music ducked 18–20dB under speech, overall loudness around −14 LUFS, effects placed a few milliseconds before the visual they accent, and silence used as a deliberate tool. A section on directing generated footage shows how to write shot prompts as pixels instead of feelings, with a fill-in template (subject, light, first frame, last frame). The review itself is a scored instrument: the slideshow-trap questions, a six-area craft scorecard (publish at 10+ of 12, no zero on hook, legibility, or sound), and findings that must name where the problem is and the concrete fix — a worked review of a 90-second explainer shows five findings, two of them blockers, and how they get resolved.

    - The numeric defaults (type sizes, loudness, ducking, interrupt cadence) are starting points for common short-form and explainer work; long-form, cinematic, or broadcast deliverables have their own norms you'll adjust to deliberately. - The review can only judge what's in the draft you show it — a script-stage review can't catch a bad render, and a render-stage review can't recover content the script never had. - It governs craft, not strategy: it won't tell you whether this video should exist or which platform to prioritize.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 24 days ago

    Permissions

    No special permissions declared or detected

    Listed24 days ago
    Updated21 days ago

    What's inside

    Frequently Asked Questions

    The rules come with numbers and reasons — caption hold time is derived from reading speed, the bottom safe zone from where platform UI actually sits, ducking levels from what phone speakers do to a mix. And the review discipline is the point: every finding must name where the problem occurs and the concrete fix, or it isn't a finding yet.

    As the standard it builds against and the final gate before publishing. The skill is written to be applied while scripting, generating, or editing — and the scored pre-publish review (10+ of 12, no zero on hook, legibility, or sound) is the part most pipelines skip.

    The most common failure of generated and templated video: stills with captions, decorative motion, the same transition repeating, and on-screen text doing the job the visuals should do. The review asks those four questions directly, and any yes is a blocker — because the result is a slideshow, not a video, no matter how clean the render is.

    New in Image & Video