linkedin-limits-warmup

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow is described as using ColdIQ MCP/API to generate LinkedIn targets (“mcp__coldiq__find_signals” / “mcp__coldiq__find_people”), which at runtime would ingest outsider-authored content from a third-party service (ColdIQ) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs running remote code via the command "COLDIQ_API_KEY= npx -y @coldiq/mcp@latest" which fetches and executes an npm package at runtime, so it's a runtime external dependency that executes remote code.