@ggts-sh/skill
v0.43.0
Published
Agent skill for the shared grammar of graphics: TypeScript, React, Svelte, portable JSON, and the ggts check/render feedback loop. Versions with all framework adapters.
Downloads
34
Maintainers
Readme
@ggts-sh/skill
The ggts agent skill: SKILL.md plus deep-dive references/ that teach a
coding agent the shared grammar of graphics — TypeScript builders, React
and Svelte components, PortableSpec JSON, and the ggts check /
ggts render feedback loop.
This package is the one published home of the skill. It versions in lock-step
with @ggts-sh/spec,
@ggts-sh/core,
@ggts-sh/compose,
@ggts-sh/react,
@ggts-sh/svelte, and
@ggts-sh/cli: a given version
number describes the spec, the renderers, and this skill as of the same
release. Pin it like any other dependency and let dependabot (or
npm-check-updates) tell you when the bundled skill in your repo is stale.
Install
npm install --save-dev --save-exact @ggts-sh/skill @ggts-sh/cliThe package root is the skill directory: SKILL.md sits next to this
README. Skill loaders key off the frontmatter (name: ggts), not the
directory name.
Use
Point your agent at the skill, or copy/symlink it into your agent's skills
directory under the name ggts:
# Claude Code
cp -R node_modules/@ggts-sh/skill .claude/skills/ggts
# pi
cp -R node_modules/@ggts-sh/skill .pi/agent/skills/ggts
# or reference it in place
node_modules/@ggts-sh/skill/SKILL.mdRe-run the copy on every version bump (a two-line postinstall or a sync
script works); the dependabot PR is the signal that the skill changed.
The skill assumes the agent can also run
@ggts-sh/cli
(ggts check and ggts render) for spec validation and headless SVG rendering — install it
in every sandbox where an agent authors specs.
Contents
SKILL.md— target selection, shared grammar, authoring workflow.references/react.mdandreferences/svelte.md— adapter instructions.references/— geoms and stats, scales and palettes, themes, interactions, composition surfaces, recipes.
Guarantees
CI checks the teaching contracts and shipped examples:
- Content contracts (
scripts/skill-content/*.test.ts,scripts/skill-package.test.ts) — inventory completeness for every geom, stat, position, theme, and color scheme; every complete JSON fence normalizes and validates; pack shape and lock-step version. Packed-consumer checks also compile complete TSX and Svelte examples from the installed skill. - Trigger / disclosure contracts (
scripts/skill-trigger.test.ts) — frontmatter description quality (the loader's selection signal), balanced positive/negative trigger fixtures underevals/trigger-cases.json, relative-link integrity, and progressive disclosure of everyreferences/file fromSKILL.md. - Held-out NL→spec evals (
tests/evals/,bun run evals) — model capability on PortableSpec authoring with deterministic graders. These do not load this skill today; agent-in-the-loop skill evals (with/without skill A/B, multi-trial trigger accuracy) are tracked as follow-up work.
evals/trigger-cases.json is not packed to npm (files is only SKILL.md +
references/). It is the living seed for future skill-loaded agent evals.
