agentleFS
hassancs91hassancs91/claude-image-generation/.claude/skills/story-html-publisher/SKILL.md

story-html-publisher

Skill87 starsChanged 4 months agoMITView on GitHub

Final step of the AI Storybook pipeline. Consolidates the scenes, images, and per-scene audio into ONE self-contained HTML storybook — a swipe/tap player with every image and audio clip embedded as base64 so the single file works offline and can be shared as-is. Reads {slug}_scenes.json, {slug}_images.json, and {slug}_audio/, then writes {slug}_story.json and {slug}.html. Use this skill whenever the user wants to build the final storybook, export to HTML, package the story, make the storybook file, generate the HTML, or wrap up the pipeline. Trigger on phrases like "build the storybook HTML", "export to HTML", "make the final file", "package the storybook", "generate the storybook", "wrap up the pipeline", or whenever the scenes + images + audio all exist and the user wants the single shareable file. Produces a single self-contained .html (images + audio embedded).

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Does NOT apply to stories that haven't been through scene-splitter + story-illustrator + story-narrator — point the user to the missing …

    line 16
  2. Distinctive:

    If anything is missing, list the gaps.

    line 40
  3. Distinctive:

    Run the build script (preferred):

    line 44
  4. Very specific:

    Does NOT generate new content (text/images/audio) — it only packages.

    line 88
  5. Distinctive:

    Does NOT upload anywhere — local file output only.

    line 89
  6. Distinctive:

    Does NOT modify upstream artifacts.

    line 90
  7. Distinctive:

    Does NOT support multiple stories per run — one story per invocation.

    line 91

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Story HTML Publisher

Takes everything the upstream skills produced and emits one self-contained HTML storybook: a swipe/tap player where each scene is one screen (illustration + narration text), with the playback controls in a fixed footer. It opens with a dedicated cover page — the first illustration shown clean, playing only the short spoken title. Every image and audio clip is embedded as a base64 data URI, so the single .html file works offline, opens with a double-click, and can be emailed or dropped on any static host with no assets folder.

This skill does no creative work — no new text, images, or audio. It only reads, validates, pairs, embeds, and writes. If something is missing upstream, it reports the gap rather than fabricating.

When this skill applies

The pipeline's earlier skills have run and the user wants the final output: "build the storybook", "export to HTML", "make the final file", "wrap up the pipeline".

Does NOT apply to stories that haven't been through scene-splitter + story-illustrator + story-narrator — point the user to the missing step.

Required inputs

In stories/{slug}/:

ArtifactProducerPath
Scenes spinescene-splitter{slug}_scenes.json
Image registrystory-illustrator{slug}_images.json
Audio clipsstory-narrator{slug}_audio/{slug}_part_NN.mp3
Title clip (optional)story-narrator{slug}_audio/title.mp3

Everything pairs by scene index: scenes[i].index = N → image from images.json.shots[] where index = N → audio {slug}_part_NN.mp3.

Workflow — four stages with one hard gate

Stage 1: Discover and validate

  1. Determine the slug (user argument → the *_scenes.json in the working dir → ask).
  2. Locate the three inputs. Confirm each scene has both an image (in images.json) and an audio file. Build a pairing table: index → text snippet → image present? → audio present?
  3. Reconstruction sanity check: the scene count in scenes.json should match the number of shots in images.json and the number of _part_NN.mp3 files.
  4. Report findings:

    Found 9 scenes, 9 images, 9 audio clips (+ title clip). All paired. Ready to build. If anything is missing, list the gaps. Don't proceed to build with gaps unless the user says "build partial" (the build script will embed what exists and leave placeholders for the rest).

Stage 2: Build the HTML

Run the build script (preferred):

python .claude/skills/story-html-publisher/assets/build_html.py {slug}

It downloads/reads every image and audio file, embeds them as data URIs, injects them into assets/player-template.html, and writes stories/{slug}/{slug}_story.json + stories/{slug}/{slug}.html. Read references/build-html.md for what it does, options, and how to do it by hand if the script can't run.

On Windows use python; on macOS/Linux use python3. Stdlib only — no installs needed.

Stage 3 — HARD GATE: review the storybook

Report:

  1. Output path: stories/{slug}/{slug}.html, its size (MB), and scene count
  2. Any warnings the build emitted (missing image/audio for a scene)
  3. A one-liner: "Open stories/{slug}/{slug}.html in your browser to read the finished storybook."

Then prompt:

Storybook built — open it and review, then reply:

  • done to finish
  • rebuild to regenerate the HTML (e.g. after swapping an image/audio file)
  • fix image N / fix audio N — if a scene's media is wrong, regenerate it upstream (illustrator/narrator), then rebuild

This is the final review surface — the user reads the actual storybook here. Don't declare success before they've had the chance to open it.

⚠️ The .html can be large (several MB) because media is embedded. It opens fine in a browser but may be slow to preview inside a chat/artifact viewer — tell the user to open it locally in a real browser.

Stage 4: Finish

On done, confirm in one line: "Storybook ready: stories/{slug}/{slug}.html ({N} scenes, {size} MB). Fully self-contained — share or host it anywhere."

What the finished storybook does

  • A dedicated cover page (slide 0): the first illustration shown clean — no overlaid title text, no scrim, no glyph — with the title in the header. It plays the optional spoken title clip, then auto-advances into scene 1.
  • One scene per screen after that: illustration + narration text, controls in a fixed footer
  • Tap the image (or the footer play button) to play/pause the scene's narration. No play glyph is drawn on the image.
  • Swipe (left = next, right = prev), arrow keys, on-screen buttons, or tap a progress dot to navigate
  • Auto-play toggle: when a scene's narration ends, advance to the next scene. Every slide is uniform (one image, one clip, advance on ended) — no title→body audio chaining
  • A "The End" screen with a "Read again" button at the finish
  • Serif body text matching the serif title, LTR English, warm storybook styling, responsive (phone + desktop), no external dependencies

What this skill does NOT do

  • Does NOT generate new content (text/images/audio) — it only packages.
  • Does NOT upload anywhere — local file output only. (The result is portable; the user can host it wherever.)
  • Does NOT modify upstream artifacts.
  • Does NOT support multiple stories per run — one story per invocation.

Reference files

  • references/build-html.md — what the build does, options, the data shape, and the by-hand fallback
  • assets/build_html.py — the build script (stdlib only)
  • assets/player-template.html — the self-contained swipe player template

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.