The Agent Skills Directory

[SAFE]: The skill documents project-specific coding standards and security practices for the firstsun-blog repository. It mandates authentication guards for administrative actions and strict input validation using Zod schemas, which are standard security best practices for web development.
[COMMAND_EXECUTION]: Instructions include running standard development scripts such as 'pnpm lint', 'pnpm build:e2e', and 'pnpm test:e2e', as well as database management commands using 'wrangler' and 'drizzle-kit'. These operations are consistent with the skill's primary purpose of guiding development workflows and maintaining project quality.
[INDIRECT_PROMPT_INJECTION]: The skill manages the processing of external data through API actions and database mutations.
Ingestion points: Data enters the system via Astro Actions in 'src/actions/index.ts' and public/admin API routes.
Boundary markers: The instructions do not explicitly define prompt boundary markers, but they mandate strict data validation using Zod schemas to delimit expected data structures.
Capability inventory: The skill has the capability to execute shell commands ('pnpm', 'wrangler') and modify project source files, which are necessary for its development role.
Sanitization: All inputs must be validated with Zod schemas, and administrative actions are protected by a 'verifyAuth' requirement to prevent unauthorized execution.
[SAFE]: Asset management guidelines explicitly advise against downloading external assets locally and favor external references, effectively mitigating the risk of importing malicious payloads via SVG or media files.

dev-conventions