cold-email

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

• Third-party content exposure detected (medium risk: 0.30). ColdIQ MCP/API calls (e.g., mcp__coldiq__fetch_page_content, mcp__coldiq__extract_post_engagement, and related “signals/enrichment” endpoints) ingest outsider-authored free text at runtime because those data are derived from web pages/social posts/prospects’ public content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill repeatedly instructs runtime calls to the ColdIQ MCP/API (and even to install/run the MCP via npx -y @coldiq/mcp@latest), which fetches external data/code that directly controls prompts and personalization — e.g., https://api.coldiq.com and the marketplace link https://coldiq.com/marketplace are used at runtime to drive the agent's behavior.