story-render
Skill4 starsChanged 3 months agoMITView on GitHub
Regenerate and render an existing story. Use to change a line of narration, swap a voice, redo one image, switch orientation, or re-render to mp4.
- Reads credentials
In this repository3 other files it gives its agents.
- story-newSkill
- story-setupSkill
- story-tellingSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
If no slug, list the folders under public/ that contain a story.json and ask which one.
- Distinctive: line 108
Read the real output duration rather than assuming the planned one
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Re-run part of an existing story without paying for the whole thing again.
Arguments
$0= slug, the folder name underpublic/$1= what changed:narration,voice,images,orientation, orrenderfor no change at all
If no slug, list the folders under public/ that contain a story.json and
ask which one.
How the skip logic works
npm run generate skips any voice-N.wav or image-N.jpg already on disk.
That is the whole cost control: delete exactly what should change, leave the
rest, and rerun. --force regenerates everything and should be rare.
Recipes
Changed a line of narration
Edit scenes[N].narration in public/<slug>/story.json, then:
rm public/<slug>/voice-N.wav
npm run generate -- --slug <slug> --only voiceThe new duration is measured and written back into story.json, so the scene
and its subtitles resize themselves. Never hand edit durationInFrames.
Changed the voice or the language
Edit voice.speaker or voice.languageCode, then regenerate all narration:
rm public/<slug>/voice-*.wav
npm run generate -- --slug <slug> --only voiceSpeaker names are lowercase. Check the catalog in
story-telling/rules/voices.md, and remember a non-Latin languageCode needs
a font that covers the script, added in src/lib/fonts.ts.
One image is wrong
Edit that image's prompt, then:
rm public/<slug>/image-7.jpg
npm run generate -- --slug <slug> --only imagesFor a face that drifted from its reference, move the likeness clause to the
front of the prompt. See story-telling/rules/images.md.
The whole look is wrong
Change style.artPrompt, delete every image, regenerate. Narration is
untouched, so this only costs the image calls:
rm public/<slug>/image-*.jpg
npm run generate -- --slug <slug> --only imagesTo iterate cheaply, set FLUX_TEXT_MODEL=fal-ai/flux-2/flash in .env, tune
the prompts, then switch back to fal-ai/flux-2-pro and regenerate.
Switching landscape to vertical, or back
Images are generated at the composition size, so all of them are now the wrong shape:
# after editing "orientation" in story.json
rm public/<slug>/image-*.jpg
npm run generate -- --slug <slug> --only imagesVoices are unaffected. Check the result in the Studio: the stat card moves to top centre and subtitles regroup to 4 words automatically.
Adding a scene
Append a scene with the next id, give its images unused ids, then
npm run generate -- --slug <slug>. Only the new assets are generated.
Just re-render
npx remotion render <CompositionId> out/<slug>.mp4 --codec h264Useful options: --concurrency 4 if memory is tight, --frames 0-120 to check
a section quickly, --scale 0.5 for a fast rough cut.
Checks before reporting
npm run typecheckclean- Read the real output duration rather than assuming the planned one
- Watch the last scene: subtitle drift shows there first
- Report duration, file size and path as measured
Example usage
/story-render big-bull narration
/story-render big-bull images
/story-render big-bull render
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

