Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →gooseworks-ai/render-myth-vs-fact
Assemble a myth-vs-fact kinetic-typography explainer video ad (≈29.5s, 9:16) from N myth/fact pairs + hook / turn / punch copy + palette + a brand end-card PNG + a VO track — a hook, 3 red-strike MYTH cards that flip to teal-check FACT cards (per-line strikethrough that crosses EVERY wrapped line), a "what actually works" turn, an optional proof reveal, a punch line, and a static end card. DETERMINISTIC assembly with ZERO AI-gen visuals — HTML hyperframes rendered frame-exact via Playwright (`wi
Passed 18 of 22 test cases with the skill, and 8 of 22 without it.
and 10 more in the full report
11 fixed7 passed either way3 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
This benchmark measured an earlier version of the skill.
One of 11 test cases the skill fixed. Same model, same prompt, with and without the skill.
Which specific Python script in the project directory is responsible for invoking Playwright browser screenshots and driving frame-by-frame HTML hyperframe rendering?
| Checked | Without skill | With skill |
|---|---|---|
| Identifies scripts/render_beats.py as the rendering script | Missed | Passed |
The instructions your model reads when the skill loads.
The free, deterministic renderer for the myth-vs-fact video ad format — the sound-off-safe kinetic-typography explainer that busts N common myths and hands the viewer a credible resolution. Red-strike MYTH cards flip to teal-check FACT cards over a VO (voice + tone are the caller's choice; the demo used a calm-authority expert), then a "what actually works" turn + an optional proof reveal + a punch line + a static brand end card.
Every on-screen word is a deterministic HTML hyperframe — no AI image/video gen, no b-roll, no character. Visuals cost $0. The only metered spend is upstream (VO + Whisper word-timestamps + an optional music bed), gated to its own capabilities. This capability OWNS the whole FREE assembly: beat-snap → render → captions → mix → burn → master. Iterate the cut for free; re-roll only the offending paid audio beat.
It ports the validated build from the Clinikally "acne myths" run — brand-neutralised,
config-driven, and portable (no /Users, no clients/; everything via --config +
--work-dir).
SOUND-OFF SAFE is the whole point: every claim is legible on-screen and the VO only reinforces it. VO-FIRST: render the VO, extract Whisper word onsets on the RENDERED audio, then snap every beat boundary + strike wipe + reveal to those onsets.
hook → 3× myth-fact (the flip triad — identical grammar so it reads as a pattern) →
turn (the "what actually works" pivot) → proof (optional actives/proof reveal, omit if
empty) → punch (full-frame closer) → end-card (the static brand PNG). Each beat carries
its role, duration, and its copy; ONE role template renders any pair.
scripts/beat_snap.py — VO-first alignment. FAL Whisper word-timestamps on the RENDERED
VO → re-snap every beat boundary to the nearest word onset. Writes
beat-manifest.json + whisper/words-flat.json into the work dir. --no-whisper keeps
the config durations un-snapped for a fully offline run.scripts/render_beats.py — the deterministic renderer. Per mg beat: pick the role
template under hyperframes/, inject the beat's copy + palette + fonts as window.BEAT,
drive window.renderAt(t) frame-by-frame via Playwright, screenshot each frame → ffmpeg
at EXACTLY the configured fps (default 25/1). The end-card beat is built from the
pre-supplied brand PNG (scale/crop + a ~0.35s fade-up) — never generated per run.test cases passed, vs 8 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.