linkedin-campaign-complete

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The required workflow calls ColdIQ MCP endpoints that can fetch and extract real third-party LinkedIn/profile/company content (e.g., fetch_page_content and post engagement), which would be ingested as readable text into the agent’s LLM context at runtime—an outsider-authored free-text source.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly calls the ColdIQ API at https://api.coldiq.com during runtime to fetch enrichment and intent signals that are then used to generate and control agent prompts and messaging, making it a required external dependency that influences instructions.