The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands for environment preflight checks (using jq, grep, ls), managing the Storybook development server (npm run, nohup), and interacting with platform-native keychain utilities (security on macOS, secret-tool on Linux, and PowerShell DPAPI on Windows) to securely handle authentication secrets.
[EXTERNAL_DOWNLOADS]: The skill references and suggests the installation of well-known development dependencies from the npm registry, including official Storybook adapters, testing runners, and the Playwright framework.
[REMOTE_CODE_EXECUTION]: To support authenticated workflows, the skill generates a temporary Playwright login script at runtime from an internal template. This script is used locally within the agent's Node.js environment to handle login forms and produce session state files, following best practices for isolated automation.
[PROMPT_INJECTION]: The skill identifies and manages an indirect prompt injection surface area by processing React component source files and interacting with rendered Storybook web content.
Ingestion points: React component source files (Phase 1), package.json (Phase 0), and external Storybook URLs (Phase 5).
Boundary markers: The workflow includes a mandatory 'Component inspection' phase where an inventory of interpreted props and variants is presented for user verification before any files are created.
Capability inventory: The skill has capabilities for file-writing (*.stories.tsx), shell execution (npx, npm), and local keychain access.
Sanitization: External data is distilled into structured inventories for user review, reducing the risk of accidental obedience to instructions embedded in the source code.

storybook