webflow-mcp:link-checker

Fail

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: CRITICALPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it ingests untrusted data from site pages and CMS items to identify links.
  • Ingestion points: Site content is retrieved via data_pages_tool:get_page_content and data_cms_tool:list_collection_items.
  • Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions when parsing the content for links.
  • Capability inventory: The skill possesses write capabilities including update_static_content and update_collection_items, allowing it to modify the site based on the processed data.
  • Sanitization: The instructions do not define any sanitization or validation steps for the extracted content before it is used to suggest or apply fixes.
  • [DATA_EXFILTRATION]: The skill interacts with arbitrary external domains to validate URL accessibility as part of its core link checking functionality.
  • Evidence: Phase 2, step 7: "Validate each link: Test URL accessibility". This requires the agent to make network requests to non-whitelisted external domains discovered within the site content.
Recommendations
  • CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Jul 21, 2026, 08:29 AM
Security Audit — agent-trust-hub — webflow-mcp:link-checker