calicastle/skills

This repository collects "Agent Skills" created by Cali Castle. The README is sparse on broader context, so a non-technical reader should know upfront: the repo does not explain what an Agent Skill is, what skills.sh is, or what Codex refers to. You can install and use the skills here, but you are expected to already understand the surrounding ecosystem. The collection currently includes two skills. The first, Signal Geometry, turns a concept into a spare abstract illustration or poster. It relies on precise geometry, muted fields, and a single clear spatial event. The second, Chalk Logic, produces a wordless white-chalk illustration on a desaturated green board. It is meant for explanatory systems, natural processes, and sparse editorial scenes, while avoiding cluttered classroom diagrams or polished digital graphics. Installation is handled through npm or pnpm, two common JavaScript package managers. The commands shown add a specific skill to your local setup. Once installed, you invoke a skill by name in a prompt, prefixing it with a dollar sign: $signal-geometry or $chalk-logic. You can ask the skill to render an image, or you can run it in prompt-only mode, which returns the prompt and composition details without generating a picture. In rendered mode, each skill returns the final image along with its exact prompt and a complete composition recipe. Chalk Logic adds a QA status to that output. If image generation tools are unavailable, Chalk Logic can route work through an isolated worker and return file links instead of images inline. Prompt-only mode works without any image tools at all. Both skills and their reference images are released under the MIT License, a permissive open-source license. The README is concise and mostly procedural, covering installation, usage, output, and requirements for each skill.