The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's cleanup script executes pnpm dev:h5 as a subprocess to trigger the automatic regeneration of src/pages.json via project plugins. It also invokes pnpm install, pnpm type-check, and pnpm lint as part of the verification workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local project files to determine execution logic and cleanup targets, creating an attack surface where a malicious project structure could influence agent behavior.
Ingestion points: The scripts/clean.js script reads and parses package.json and vite.config.ts from the project directory.
Boundary markers: None identified; the script processes file contents directly without delimiters or explicit safety instructions regarding the content of these files.
Capability inventory: The skill has capabilities for recursive file deletion (fs.rmSync), file modification (fs.writeFileSync), and arbitrary command execution (spawn).
Sanitization: The script uses JSON parsing and basic string/line filtering to process inputs, which does not provide robust protection against malicious configuration values.

starter-cleaner