The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines instructions for the agent to configure projects based on external files like netlify.toml, _headers, and _redirects. These files represent an ingestion point for untrusted data if the agent processes a shared or external repository.
Ingestion points: Configuration files (netlify.toml, _headers, _redirects) are parsed to define site behavior.
Boundary markers: The documentation includes "Footguns" and "House rules" sections that explicitly warn against insecure configurations, such as committing .env files or exposing secrets in client-prefixed variables.
Capability inventory: The skill documents shell command execution via the command property and the usage of build plugins and environment variable injection.
Sanitization: The skill emphasizes the use of the Netlify Secrets Controller for write-only secret management and the use of the official CLI for secure variable handling.
[DYNAMIC_EXECUTION]: The skill demonstrates the use of sed within build commands to dynamically inject values into netlify.toml at runtime. This practice involves modifying configuration files during the build process, which is a standard pattern for environment-specific customization on the Netlify platform.
[SAFE]: The skill is authored by the official vendor (Netlify) and primarily focuses on educational content and best practices for their service. It uses standard placeholders for examples and proactively advises users on how to avoid common security pitfalls like credential exposure and improper environment scoping.

netlify-config