The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill primarily serves as a documentation reference for Webflow developers.
[COMMAND_EXECUTION]: The skill includes a local utility script scripts/search_references.py. Analysis of the script confirms it is safe; it performs text-based searching using standard libraries (re, pathlib) and does not use dangerous functions like eval(), exec(), or subprocess.
[PROMPT_INJECTION]: The skill implements a defensive mechanism against prompt injection. The scripts/search_references.py file contains a regex _INJECTION_RE and a sanitize_body function specifically designed to strip out common prompt injection markers (e.g., 'ignore previous', 'system:', 'override') before displaying content to the agent.
[EXTERNAL_DOWNLOADS]: The documentation references standard developer tools such as npx create-webflow-extension and @webflow/webflow-cli. These are recognized as legitimate utilities within the Webflow development ecosystem. External URLs point to official Webflow documentation and the well-known placeholder service picsum.photos.
[DATA_EXPOSURE]: The skill uses safe placeholders for sensitive data (e.g., YOUR_CLIENT_ID, YOUR_SCOPES) and explicitly warns against committing Client Secrets to version control in references/register-app.md.

webflow-designer-api