Skip to main content

reveal-shot-prompts

Why It Exists​

This skill exists to formalize a pipeline the user was already running by hand: nano-banana ai for a hero product image, an explosion-shot prompt fed to whiskailabs.org for a deconstructed still, a transition-flow prompt fed to labs.google/fx/tools/flow (Veo 3.1) for a reveal video, and ezgif.com to pull 28-30 still frames back out. The originating plan (.omc/plans/reveal-shot-scrollytelling-skills.md) records this as a locked interview decision, not a default: only steps 1-2 (the two prompt templates) get automated; steps 3-4 (Flow, ezgif) stay "manual checklist only — no browser automation," and a related step-5 skill only ever produces a filled prompt, never scaffolds a site.

The introducing commit (36e1d0e, feat(skills): add reveal-shot-prompts and scrollytelling-frames-site skills) frames both new skills as turning "the manual product-reveal-video-to-scrollytelling-site workflow into two prompt-templating skills," explicit that neither "calls a generation API, automates a browser, or scaffolds code." That commit also wired the skill into the repo's package/publish pipeline per docs/skill-addition-runbook.md, regenerating the marketplace manifest from 19 to 21 skills — landed first as unpublished, then enabled for publish in a follow-up commit (b585541) the same session (docs/sessions/2026-08-17-docs-site-and-skills/SUMMARY.md).

Keeping the browser steps manual was a deliberate scope cut, not a limitation to fix later: the plan's "Out of Scope" section calls out "no browser automation of labs.google/fx/tools/flow or ezgif.com" and "no actual site scaffolding/code generation for step 5" as explicit, interview-derived boundaries.

What It Does​

reveal-shot-prompts fills two prompt templates from user-supplied inputs (product name, product type, key internal parts, background color, primary elements, energy effect): an explosion-shot prompt for a deconstructed engineering visualization, and a transition-flow prompt for a cinematic assembled-to-exploded video transition. Each filled prompt is printed as a complete, ready-to-paste code block for nano-banana ai.

Around those two prompts, it also prints two checklists verbatim for the user to run by hand: configuring a labs.google/fx/tools/flow video project (Video, Frames, 16:9, Veo 3.1 Quality) and extracting 28-30 still frames from the resulting video at ezgif.com/video-to-jpg. The skill never calls an image- or video-generation API and never automates a browser — every whiskailabs.org, Flow, and ezgif.com step is performed by the user.

How To Use It​

Triggers on: "explosion shot prompt", "product reveal video", "whisk ai deconstructed prompt", "flow transition prompt", "nano-banana", "deconstructed engineering visualization", "exploded product shot", "assembled to exploded transition".

skills add git@github.com:catesandrew/next-starters.git --skill skills/reveal-shot-prompts -g
npm install @next-starters/skill-reveal-shot-prompts
/plugin marketplace add catesandrew/next-starters
/plugin install reveal-shot-prompts@next-starters

Gotchas & Invariants​

  • The skill only produces prompt text and checklists — it never calls an image- or video-generation API and never automates a browser.
  • Step 3's Flow settings are "non-negotiable": Video (not Image), Frames (not Ingredients), 16:9 (not 9:16), and Veo 3.1 Quality (not Lite/Fast) must all be selected exactly as printed.
  • Step 4 must extract 28-30 frames at ezgif.com/video-to-jpg — not an arbitrary count.
  • `${input:product_name}` fills both `[PROJECT NAME]` and `[PRODUCT NAME]` in the transition-flow template — the two placeholders are the same value reused, not two independent inputs.
  • `${input:background_color}` defaults to #0A0A0A if the user doesn't supply one; every other input is required with no default.
  • The explosion-shot prompt is handed off to whiskailabs.org (not labs.google/fx) for the deconstructed still — the transition-flow prompt is what goes to Flow, alongside both the hero image and that whisk-generated still.
  • This skill only covers steps 1-4 of the 5-step pipeline; step 5 (turning extracted frames into a site) is out of scope by design and belongs to the sibling scrollytelling-frames-site skill.
  • scrollytelling-frames-site — the downstream skill that fills an antigravity prompt turning this skill's ezgif-extracted frame sequence into a scroll-linked site.

Sourced from: skills/reveal-shot-prompts/metadata.json, skills/reveal-shot-prompts/SKILL.md, .omc/plans/reveal-shot-scrollytelling-skills.md (local research artifact, not tracked in this repo), docs/sessions/2026-08-17-docs-site-and-skills/SUMMARY.md, docs/sessions/2026-08-17-docs-site-and-skills/README.md, docs/sessions/2026-08-17-docs-site-and-skills/FOLLOWUPS.md, git log --follow -- skills/reveal-shot-prompts (commit 36e1d0e)