The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external content which constitutes a potential attack surface for indirect instructions.
Ingestion points: Uses Read, Glob, and Grep to analyze local repository files (e.g., webflow.json, package locks) and WebFetch to retrieve content from developers.webflow.com.
Boundary markers: The skill includes a dedicated 'Approval Boundaries' section that mandates explicit user confirmation before any live mutation (create, update, delete, etc.) or production publication.
Capability inventory: The skill possesses Write and Edit capabilities for modifying local files and uses the Webflow CLI for cloud deployments.
Sanitization: Employs mandatory dry-runs, pinned versions, and strict separation between read-only PR checks and mutation-capable workflows.
[EXTERNAL_DOWNLOADS]: The skill fetches configuration and documentation from well-known official sources.
Evidence: WebFetch is used to access developers.webflow.com and related subdomains for Data API and CLI references.
Risk Assessment: These references target official Webflow developer resources, which are recognized as safe and well-known services.

webflow-ci-integration