workorai

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

• Third-party content exposure detected (high risk: 0.70). The required workflow connects to a remote MCP server and then uses its tools (e.g., employer.search_candidates_for_job / employer.get_applicant_transcript) which can return outsider-authored free-text (candidate transcripts, rationales, etc.) that the agent will ingest into LLM context via the MCP tool responses.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires adding and calling the remote MCP server at https://workorai.com/mcp at runtime (see the claude mcp command and "All operations go through the remote WorkorAI MCP server" note), which means external content from that URL controls the agent's available tools and runtime behavior.