← marketplace
creativedesignOfficialsha:297c84a947f6060bmanual

algorithmic-art

Use when a user asks for generative or algorithmic art — flow fields, particle systems, noise-driven visuals — so the agent designs an aesthetic philosophy then ships it as interactive p5.js.

Install confidence
curl --create-dirs -fsSL https://skillmake.xyz/i/algorithmic-art -o ~/.claude/skills/algorithmic-art/SKILL.md
Pinned content
sha:297c84a947f6060b
Generated with
manual
Source
github.com

The file served at /api/marketplace/algorithmic-art-297c84a9/raw matches this hash. Inspect before install, then copy the command.

3,357 chars · ~839 tokens
---
name: algorithmic-art
description: Use when a user asks for generative or algorithmic art — flow fields, particle systems, noise-driven visuals — so the agent designs an aesthetic philosophy then ships it as interactive p5.js.
source: https://github.com/anthropics/skills/tree/main/skills/algorithmic-art
generated: 2026-07-02T18:41:15.919Z
category: design
audience: creative
---

## When to use

- Turning a loose request like 'make me some generative art' into a coherent, tunable p5.js sketch
- Building flow fields, particle systems, or noise-driven visuals with seeded reproducibility
- Producing a single self-contained interactive HTML artifact with parameter sliders and seed navigation
- Encoding a subtle conceptual reference into generative parameters without copying an existing artist's work

## Key concepts

### Two-step: philosophy then code

The skill first writes an 'algorithmic philosophy' (a 4-6 paragraph manifesto naming a generative movement), then expresses it as p5.js. The art is meant to be ~90% algorithm, ~10% essential parameters.

### Seeded randomness (Art Blocks pattern)

randomSeed(seed) and noiseSeed(seed) are set from one value so the same seed always reproduces identical output. Seed navigation (prev/next/random/jump) lets users explore variations from one algorithm.

### Process over product

Beauty is meant to emerge from the algorithm's execution, not a static frame. The skill steers toward living systems — accumulation, feedback, emergence — rather than images with random noise sprinkled on.

### viewer.html template is the starting point

templates/viewer.html holds the fixed layout, Anthropic branding (Poppins/Lora, light theme), seed controls, and action buttons. You replace only the algorithm and parameter controls, never rebuild the shell.

### Deduced conceptual seed

A subtle, niche reference from the user's request is woven invisibly into parameters and behaviors — recognizable to those who know it, while everyone else just sees a masterful composition.

## API reference

```
npx skills add anthropics/skills --skill algorithmic-art
```

Install the algorithmic art (p5.js generative art) skill.

```
npx skills add anthropics/skills --skill algorithmic-art
```

```
randomSeed(seed); noiseSeed(seed);
```

Fix the seed so a given seed always reproduces the same artwork.

```
let seed = 12345;
randomSeed(seed);
noiseSeed(seed);
```

## Gotchas

- Read templates/viewer.html first and build on it — creating HTML from scratch loses the fixed layout, branding, and seed controls
- Keep the Anthropic branding (Poppins/Lora fonts, light theme, gradient backdrop); don't invent custom color schemes or dark themes for the UI
- Same seed must always produce identical output — non-deterministic code breaks the whole reproducibility model
- Don't copy the flow-field example in the template; build what the philosophy demands rather than reusing the sample algorithm
- Create original algorithmic art, not reproductions of existing artists' work, to avoid copyright violations
- Ship one self-contained HTML artifact with everything inline (only p5.js from CDN) — no external files or separate .js imports

---
Generated by SkillMake from https://github.com/anthropics/skills/tree/main/skills/algorithmic-art on 2026-07-02T18:41:15.919Z.
Verify against source before relying on details.

File: ~/.claude/skills/algorithmic-art/SKILL.md