The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the creation of pages and API routes that ingest untrusted data, which establishes an indirect prompt injection surface. Monitoring this surface is a standard requirement for web development tools.
Ingestion points: Custom routes process user-controlled input via Astro.params and request.json(), and examples demonstrate fetching data from external APIs.
Boundary markers: Code templates do not explicitly include delimiters for untrusted data, although this is typical for the intended code-generation use case.
Capability inventory: The skill enables the generation of scripts capable of network operations (fetch) and file system access within the project scope.
Sanitization: The instructions proactively recommend using validation libraries like Zod to sanitize incoming data in API routes.
[EXTERNAL_DOWNLOADS]: Example code includes network requests to api.github.com to fetch metadata for the vendor's own repositories. These operations target a well-known service and are consistent with the skill's stated purpose.
[COMMAND_EXECUTION]: The primary function of the skill is to generate executable .astro and .ts files. This is the intended behavior of a development-oriented skill and is performed within the local project environment.

custom-pages-and-apis