# Published creative AI skills

Checked September 28, 2026. The footer screenshot is an unaltered capture of the official Anthropic GitHub page.

- [Anthropic Algorithmic Art](https://github.com/anthropics/skills/blob/main/skills/algorithmic-art/SKILL.md): p5.js generative art, seeded randomness and interactive parameters. Anthropic's [marketplace manifest](https://github.com/anthropics/skills/blob/main/.claude-plugin/marketplace.json) includes it in example-skills. See the [official repository](https://github.com/anthropics/skills#try-in-claude) for installation instructions. The publicly available skill is concrete evidence of the methods its instructions discuss, not evidence that particular artists' work trained an underlying model.

The project studies visible algorithmic methods and their public histories. It cannot reconstruct a model's private training dataset or prove influence from appearance alone. Resource links imply neither endorsement nor a training-data inclusion claim.

## A vocabulary for making algorithmic art

The footer pairs the existing skill’s capabilities with our suggested teaching workflow: observable quality → named method → adjustable control → specific, sourced contribution. A worked brief uses Perlin-noise flow fields and fixed-seed comparisons. It links to the interactive vocabulary so artists, students, teachers and their agents can develop and test more precise requests.

The OpenAI image-generation card was removed to keep this resource focused on algorithmic art. This is an editorial focus, not a claim that the Anthropic skill lacks vocabulary or that an improvement in outputs has been experimentally measured. The external skill runs in the visitor’s own tool; the site supplies exercises and research. Established method names and credited or chosen person names preserve attribution without assigning a whole aesthetic to one inventor.