45ck/content-machine/scene-by-scene-faceless-pipeline
Claude Code and agent skill by 45ck/content-machine
- community
Checking that the SKILL.md file is still in the repository.
Build a faceless short as a deterministic scene queue where each scene gets its own narration, caption timings, source clip, and final assembly step instead of treating the whole video as one undifferentiated blob.
Install
npx vskill@latest install 45ck/content-machine/scene-by-scene-faceless-pipelineSKILL.md
Scene-By-Scene Faceless Pipeline
Use When
- A faceless short needs a clean stage contract per scene.
- You want a deterministic baseline pipeline before adding heavier
editorial logic.
- The run benefits from queueing and per-scene artifact creation rather
than giant whole-video mutation.
Core Rule
- Build the short one scene at a time, then assemble it.
- Every scene should earn its own audio, caption timing, and visual
selection.
Inputs
- ordered scene list with narration text and visual search terms
- voice choice and music mood
- target orientation and caption style
Outputs
- per-scene narration audio
- per-scene caption timing
- per-scene visual clip selection
- assembled short with optional music bed
Workflow
- Queue the job so only one full render mutates shared output state at
a time.
- For each scene:
generate narration, normalize audio, transcribe for caption timing, select one fitting visual clip, and record the scene artifact bundle.
- Exclude already-used source clips when selecting later visuals.
- Sum the true scene durations before final render.
- Choose music only after the total duration is known.
- Render the final short from the scene bundles, then clean temporary
files.
Good Pattern
- each scene produces a small auditable artifact bundle
- audio timing drives scene duration
- source clip reuse is actively avoided
- the final render is assembly, not discovery
Bad Pattern
- one giant prompt that decides script, timing, visuals, and music in a
single opaque step
- picking music before total duration exists
- reusing the same stock clip across adjacent scenes by accident
- hiding per-scene failures inside one final render crash
Pair With
- Use with `token-level-caption-timestamps`
when the caption system needs word-level timing.
- Use with `retry-with-cache` or
`partial-regeneration` when only some scenes need reruns.
Aggregated From
gyoridavid/short-video-makerShortCreator.ts- deterministic faceless pipeline patterns from short generators
Validation Checklist
- Each scene has separate audio, timing, and visual artifacts.
- Real scene durations are summed before final assembly.
- Source clip reuse is controlled intentionally.
- Final render quality is inspectable scene by scene, not just at the
very end.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingexperiments/agent-evals/final-showcase-project/.content-machine/skills/scene-by-scene-faceless-pipeline/SKILL.md
Popularity
- Repository stars
- 16
- Repository forks
- 3
- 7-day trend score
- 19
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/45ck/content-machine/scene-by-scene-faceless-pipeline)