Each file here defines the prompt for one isolated sub-agent. The orchestrator (/component-create) invokes them via the Task tool, one phase at a time. A sub-agent sees only the prompt the orchestrator hands it — no conversation history, no other sub-agent's output, no chat memory. This is the structural guarantee…
Isolated sub-agent that re-reads every file written this run, parses props/events/slots/animations independently, and diffs against the spec. Cross-checks against validate-spec-compliance.mjs.
Isolated sub-agent that runs the Figma MCP read-only verbs on the target frame and emits a structured token JSON. Never decides anything, never maps to DESIGN.md.
Isolated sub-agent that writes the component .vue files + updates packages/webkit/package.json#exports. Strictly spec-bound; every prop/event/slot must come from the spec.
Isolated sub-agent that validates a .specs/ .md against schema, body rules, and the Constraints block. On pass, computes sha256(body), writes checksum, and flips status: draft → approved.
Runs and explains the @aziontech/webkit adoption flow — npx @aziontech/webkit init. Sets up deps, lint configs, pre-commit, the webkit MCP, and the Claude Code bundle in this project, idempotently.
Measures how much of an app already renders through @aziontech/webkit versus hand-rolled UI, and produces a prioritized backlog of custom components that have a webkit equivalent to adopt.
Answers "which @aziontech/webkit component should I use, and how?" Grounds every answer in the installed catalog and the webkit MCP — never invents a component or an import path.
Reviews a diff for correct and performant @aziontech/webkit usage — flat imports, theme tokens, tree-shaking, reuse over reinvention. The local (non-CI) design-system review.
Runs the visual and behavioral QA pass on a @aziontech/webkit app — builds and serves it, drives real routes with headless Chromium, and reports only what it observed across themes, console, a11y, states, and interaction.