The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions proactively advise against accessing sensitive local credential files, specifically mentioning ~/Library/Preferences/netlify/config.json as a location that should not be used for token retrieval.
[EXTERNAL_DOWNLOADS]: The skill contains references to Netlify's official documentation and API domains (docs.netlify.com, open-api.netlify.com), which are well-known and trusted services.
[INDIRECT_PROMPT_INJECTION]: The skill describes handling user-provided form data, which is an external data ingestion surface.
Ingestion points: Web form submissions processed by Netlify.
Boundary markers: Leverages spam filtering (Akismet) and honeypot techniques to identify malicious input.
Capability inventory: Provides instructions for form creation and API-based data retrieval.
Sanitization: The documentation confirms that Netlify sanitizes submitted data by escaping HTML tags such as <script>.

netlify-forms