Examples
May 3, 2026 · View on GitHub
Use Archetypes first if you are choosing what kind of short to make. Use Quality And Review before promoting a render as ready. Use the Graphics Archetype Remake Plan when improving non-Reddit motion/card-heavy examples.
Showcase
- Reddit Post Over Gameplay — default Reddit story mode: full-screen gameplay, Reddit opener card, captions, and no random clips
- Showcase Gallery — fast visual map of demos, skills, maturity, and what each preview proves
- Example Pages — categorized index of the example docs folder
Status Summary
Demo-backed status labels are checked against
docs/demo/manifest.json by
npm run showcase-status:check.
| Lane | Status | Demo |
|---|---|---|
reddit-post-over-gameplay | golden showcase | demo-9 |
reddit-story-split-screen | workflow; rebuild demo | archived, not public showcase |
stock-b-roll-explainer | showcase candidate | demo-10 |
text-thread-reveal | showcase candidate | demo-11 |
saas-problem-solution | showcase candidate | demo-12 |
fast-facts-countdown | showcase candidate | demo-13 |
motion-card-lesson | showcase candidate | demo-14 |
faceless-mixed-short | showcase candidate | demo-15 |
gameplay-confession-split | showcase candidate | demo-16 |
micro-doc-breakdown | proving candidate | demo-17 |
content-machine-reddit-gameplay-remix | supporting showcase candidate | demo-18 |
content-machine-motion-cards | supporting showcase candidate | demo-19 |
procedural-gameplay-backgrounds | supporting showcase candidate | demo-20 |
demo-20 is now a 1080x1920 supporting showcase with audit and
provenance evidence. Keep it framed as an additive 3D/procedural
treatment, not the default content archetype.
Prompt Examples For Agent Harnesses
Use these inside Claude Code, Codex CLI, Cursor, or another harness after installing the pack:
- "Run Content Machine doctor and tell me what is missing before generation."
- "Make a faceless stock-footage explainer about
{topic}and review the final MP4 before calling it ready." - "Analyze
{referenceVideoPath}as a winner, then generate a new short with the same pacing, not the same footage." - "Turn
{longformVideo}into candidate shorts withlongform-to-shorts; show the candidate plan before rendering." - "Use
reddit-post-over-gameplayfor this story. Keep gameplay full-screen and show the Reddit opener card for 3-5 seconds."
Secondary Real Renders
- Stock Footage Edutainment — faceless-information showcase candidate with stock-style motion, narration, captions, and fast transition pulses
- Text Message Drama — message story showcase candidate with staged chat cards over gameplay support
- SaaS Problem Solution — product-card showcase candidate with pain, proof, and CTA beats
- Facts Listicle — countdown facts showcase candidate with real voiceover, timed captions, card resets, and publish-prep validation
- Motion Card Lesson — card-native educational showcase candidate with prompt, step, payoff, captions, and publish-prep validation
- Faceless Mixed Short — mixed-source explainer candidate with voiceover, music bed, captions, diagrams, UI cards, and practical tip beats
- Subway Confession Story — gameplay-backed storytime showcase candidate with top support footage, bottom Subway Surfers-style gameplay, and midpoint captions
- Micro-Doc Breakdown — documentary-style proving candidate with archival cards, evidence inserts, narration, captions, and a base publish-prep pass
- Content Machine Self-Demo Short — repo-explainer workflow with tracked Reddit/gameplay and motion-card preview variants
- Procedural Gameplay Backgrounds — additive 3D/procedural gameplay treatment for caption-clean retention motion
- Gemini Image-Led Shorts — real rendered MP4 gallery from the image-led path; useful as a supporting showcase, not the flagship lane
Workflows And Walkthroughs
These are runnable technique docs. Some use provider keys or older surfaces and should not be confused with flagship examples. Legacy CLI walkthroughs are marked in their pages; prefer harness-first skills and flows for new docs.
Content Machine self-demo
Use this when you want a short that explains the repo with its own
generate-short path:
cat skills/generate-short/examples/content-machine-self-demo.request.json | \
node --import tsx scripts/harness/generate-short.ts
Each completed generate-short run now emits
provenance/asset-ledger.json; publish-prep reviews that ledger by
default, so example promotion should include both the asset ledger and
the publish-prep provenance.json result.
If provider keys or rate limits are not available, run the no-key smoke
path instead. See
Content Machine Self-Demo Short.
Tracked supporting previews are
demo-18
and demo-19.
Listicle from a trending topic
cat <<'JSON' | node --import tsx scripts/harness/brief-to-script.ts
{
"topic": "5 things every dev should know about Docker",
"archetype": "listicle",
"outputPath": "output/examples/docker-tips/script.json"
}
JSON
Side-by-side comparison
cat <<'JSON' | node --import tsx scripts/harness/run-flow.ts
{
"flow": "generate-short",
"runId": "redis-vs-pg",
"input": {
"topic": "Redis vs PostgreSQL for caching",
"audio": { "voice": "af_heart" },
"visuals": { "provider": "pexels", "orientation": "portrait" },
"render": { "fps": 30, "downloadAssets": true }
}
}
JSON
Research-driven news video
cat skills/reverse-engineer-winner/examples/request.json | \
node --import tsx scripts/harness/reverse-engineer-winner.ts
AI-generated visuals (no stock footage)
Requires GOOGLE_API_KEY and NanoBanana config:
cat <<'JSON' | node --import tsx scripts/harness/run-flow.ts
{
"flow": "generate-short",
"runId": "nanobanana",
"input": {
"topic": "5 tips for writing better TypeScript",
"visuals": { "provider": "nanobanana", "orientation": "portrait" }
}
}
JSON
See NanoBanana + Ken Burns example for full config.
Additional Walkthroughs
- Latest News Listicle — research → script → audio → visuals → video
- Split-Screen Gameplay — gameplay on one half, stock B-roll on the other
- Procedural Gameplay Backgrounds — 3D/procedural gameplay-like backgrounds as an additive visual treatment
- Complex Plane Rotation — custom Remotion template with drawn diagrams
- NanoBanana + Ken Burns — AI-generated images with motion effects
- NanoBanana + Veo — AI images turned into video clips
- Import Render Templates — using and creating custom templates
Proving Artifacts
Internal proving bundles live under experiments/ and are useful for
honest maturity tracking, not for first-time showcase docs.
- Proving Wave 1 — current best non-flagship candidates plus their review failures
- Proving Wave 3 — one-at-a-time archetype proving with embedded demo MP4s
Runnable Examples
The fixtures/examples/ directory contains the repo's self-contained
example packages you can run from source:
ls fixtures/examples/
Demo Gallery
See generated output samples and maturity labels in
docs/demo/.