Cali Castle Agent Skills Guide
A practical guide to Cali Castle Agent Skills, including Signal Geometry, Chalk Logic, installation commands, usage patterns, and limits.
Key takeaways#
- Cali Castle Agent Skills is a public collection of practical agent skills, currently including Signal Geometry and Chalk Logic.
- The skills are installed from the
CaliCastle/skillsGitHub repository with theskillsCLI through npm or pnpm. - Signal Geometry turns a concept into sparse abstract posters and illustrations.
- Chalk Logic turns a concept into quiet white-chalk illustrations on a desaturated green board.
Overview#
Cali Castle Agent Skills is a focused GitHub repository for reusable agent skills. The repository documents two skills: Signal Geometry and Chalk Logic. Each skill gives an AI agent a constrained creative pattern instead of asking the model to invent a visual style from scratch every time.
Signal Geometry is designed for sparse abstract illustrations and posters. Its README describes precise geometry, quiet matte fields, restrained contrast, and one legible spatial event. Chalk Logic is built for explanatory chalk-board style illustrations and supports systems, natural processes, observed objects, and simple editorial scenes.
Installation#
Install Signal Geometry with npm:
npx skills add CaliCastle/skills --skill signal-geometry
Install it with pnpm:
pnpm dlx skills add CaliCastle/skills --skill signal-geometry
Install Chalk Logic by changing the skill name:
npx skills add CaliCastle/skills --skill chalk-logic
Usage#
Invoke the skill explicitly in the agent prompt. For example, ask: Use $signal-geometry to create a dark 4:5 poster about a weak signal becoming stable through repeated filtering. No text. Prompt-only mode can return the composition recipe without rendering an image. Rendered mode needs image generation and image inspection capabilities. The repository notes that the Codex integration is explicit-only by design, so users should invoke the skill by name.
When to use it#
Use these skills when you want repeatable visual direction from an AI agent. They are not broad design suites. They are small, opinionated prompt systems that help an agent stay inside a clear visual grammar. That makes them useful for blog illustrations, concept sketches, internal docs, educational explainers, or creative experiments where consistency matters more than unlimited style range.