Open-source project
elder-plinius/FABLE-SHOWCASE avatar
elder-plinius/FABLE-SHOWCASE

FABLE-SHOWCASE: 57 Dependency-Free Browser Demos Built by Claude Fable 5 Agents

an experiment in pushing the limits of one-shot prompting

112 stars41 forksHTMLLicense varies

At a glance

What is it?
A public repository of 57 standalone HTML creative demos produced by Claude Fable 5 subagents running in parallel in a single day from one prompt. Each file opens directly in any browser with no build step, and the repository documents what a large multi-agent pipeline can generate in one sprint.
Who is it for?
Developers studying multi-agent AI workflows will find the most value here: the repository shows a concrete output from a fleet of roughly 275 subagent runs across a design, build, enhance, and QA pipeline. Creative coders looking for unusual browser experiments can open any of the 57 demos without touching a terminal.
Can I use it commercially?
Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
Is it still maintained?
Yes. The repository last received commits 84 days ago.
What is it written in?
Mainly HTML, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on September 17, 2026, and from our analysis. They are not legal advice.

Editorial analysis

What the Repository Contains and Who It Is For

This is an open-source repository of 57 self-contained creative web experiments, all generated by Claude Fable 5 agents in a single day from a single prompt. The author describes it as an answer to the question of what happens when an entire week of API credits is spent in one session. The result is a browsable gallery of generative art, physics simulations, procedural worlds, musical instruments, and interactive toys, each delivered as one HTML file with zero external dependencies.

The primary audience is developers and researchers interested in multi-agent AI workflows. For them, the repository is a concrete artifact showing what roughly 275 subagent runs can produce when structured around a multi-stage pipeline. A secondary audience is creative coders who want a collection of unusual browser experiments to study or adapt, since every file runs as-is without a package manager or build tool.

Six Categories and What They Cover

The 57 demos are grouped into six categories. Simulation and Life includes boids murmuration, Primordia particle-life, Physarum slime mold, a reef, a tidepool ecosystem, and an L-system garden. Generative Art covers Fractal Voyager, Oracle (an I-Ching visualizer), Arcana (tarot), Glassworks stained glass, Loom geometry, Shodou calligraphy, and a Mandala generator. Physics and Motion offers an n-body orbital sandbox, Event Horizon with black-hole gravitational lensing, Chladni cymatics, Clockwork gears, and a Wavelab ripple tank.

Worlds and Skies contains Aurora, an infinite gothic cathedral, Halcyon (a cinematic ocean), Metropolis, Terra Minor (a pocket planet), and Hexcrawl maps. Sound and Music includes Prism (a light synthesizer), Harmonia (a playable circle of fifths), Midnight Radio, and Swarm Symphony. Play and Curiosities covers Nebula Tilt pinball, a falling-sand terrarium, a maze observatory, Noir Terminal, Chronos (a clock zoo), and a zen rock garden.

The breadth of categories reflects the original prompt's instruction to build creative projects of all sorts in parallel. Demos within a single category vary considerably in visual complexity because each was an independent subagent task with its own design phase.

How the Multi-Agent Pipeline Operated

The generation started with a single prompt in Claude Code asking for parallel, AI-powered creative demos using Fable 5's full capabilities. That prompt fanned out to approximately 275 Claude Fable 5 subagent runs across background multi-agent workflows. Each demo passed through four stages in sequence: design, build, enhance, and QA verification.

The pipeline ran until the week's API credits were exhausted. Thumbnails were captured by rendering each completed demo in a headless browser, and the gallery embeds those captures as data URIs so the gallery itself requires no network access. The README does not document the per-demo prompts or the precise duration of individual subagent runs beyond the single-day constraint.

Repository Structure and Opening the Demos

The top level of the repository contains one `index.html` that serves as the gallery, a `thumbs/` directory with the captured preview images, and one subdirectory per demo, each named after the demo and each containing its own `index.html`. The gallery is searchable and supports category filtering; thumbnails are embedded inline so it works offline.

To open the gallery, clone the repository and open `index.html` in any browser. To open a specific demo, navigate directly to `<demo-name>/index.html`. No `npm install`, no Node.js, no CDN is required. Every demo is self-contained: no network requests are made at runtime, and no build step is needed. This makes the collection portable to environments where network access is restricted or where running downloaded executables is blocked.

The Zero-Dependency Constraint and Its Trade-offs

Each demo being a single HTML file with no dependencies is a deliberate constraint. The README specifies no CDN, no frameworks, no network requests. All rendering logic, asset data, and styling are inlined into one file. For generative art and physics simulations this is feasible because they rely on the Canvas API or WebGL, both available in every modern browser without loading anything external.

The trade-off is file size and maintainability. A self-contained demo that embeds its thumbnail as a data URI and inlines all its logic can grow large. There is also no shared code between demos, so improvements to one simulation do not propagate to similar ones. The constraint suits the original purpose of a one-day creative sprint, but it would be a poor fit for a collaborative project intended to evolve over time.

Limitations and Missing Documentation

Several practical limitations are worth knowing before using this repository. First, there is no license file or license declaration in the README. Without a clear license, default copyright terms apply to individual demo files, which means reusing them in commercial or redistributed work is legally uncertain.

Second, the README documents the generation pipeline at a high level but provides no per-demo documentation. There is no explanation of the algorithm behind any individual demo, no known-issues list, and no contribution guide. Third, quality across 57 demos generated in parallel will vary. The QA stage was automated, and not every edge case in every demo may have been caught. The repository has no GitHub releases and no versioning scheme, so there is no clear way to track changes to individual demos over time. Anyone who needs predictable, versioned output from a generative art library should look at a maintained project such as p5.js, which provides a stable API and comprehensive documentation.

Editorial conclusion

Developers studying multi-agent AI workflows will find the most value here: the repository shows a concrete output from a fleet of roughly 275 subagent runs across a design, build, enhance, and QA pipeline. Creative coders looking for unusual browser experiments can open any of the 57 demos without touching a terminal. Anyone wanting a production-quality component library should look elsewhere. The repository carries no license, so confirm the terms before reusing individual demo files in commercial work.

Frequently asked questions

Do the demos in this repository work offline?

Yes. The README states that every demo is a single index.html file with zero dependencies and no network requests. The gallery also embeds its thumbnails as data URIs, so both the gallery and each individual demo work without an internet connection.

How were the gallery thumbnails generated?

The README states that thumbnails were captured by rendering each completed demo in a headless browser. They are embedded as data URIs directly inside the main gallery index.html, which is why the gallery itself has no external dependencies.

Is there a license for the individual demo files?

The repository carries no license declaration. The README does not specify a license for the demo files. Under default copyright terms, the code is not freely reusable without explicit permission from the author.

Official sources

  1. Dépôt GitHub elder-plinius/FABLE-SHOWCASE
  2. Galerie index.html du dépôt
  3. README du projet
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/elder-plinius-fable-showcase.svg)](https://hysenlabs.com/projects/elder-plinius-fable-showcase)