workos-to-descope

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

• Third-party content exposure detected (high risk: 0.85). Part 1/2/3 rely on docs_search/docs_ask_question (MCP) to fetch Descope documentation text at runtime, which is outsider-authored free text from a public docs source; this text is then ingested 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). This skill explicitly requires and queries the Descope MCP server at runtime to verify SDK method names and control generation (see "Descope MCP is not installed" and server URL https://mcp.descope.com / docs page https://docs.descope.com/mcp/mcp-server), so the external MCP endpoint is a runtime dependency that directly influences prompts and code the agent emits.