The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents patterns for handling HTTP requests within serverless and edge functions. These functions naturally ingest untrusted data from external sources.
Ingestion points: Functions in netlify/functions/ and netlify/edge-functions/ process data from req.json(), req.url, and context.params.
Boundary markers: The instructions provide examples of method validation (e.g., checking req.method) and response status handling.
Capability inventory: The functions utilize fetch for network requests and Netlify Blobs for key-value storage access.
Sanitization: Examples demonstrate validation practices, such as file size limits in uploads.mts and usage of typed Context for geo-data and cookies.
[EXTERNAL_DOWNLOADS]: The skill references and integrates official plugins and libraries from Netlify, such as @netlify/plugin-lighthouse and @netlify/functions. These are well-known resources provided by the platform vendor.
[CREDENTIALS_UNSAFE]: The skill explicitly warns against storing secrets in netlify.toml and provides secure alternatives using the Netlify UI, CLI, or environment variable management.

infra-platform-netlify