pingone-to-descope

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md instructs the agent to read the user’s local repository contents (Step 1 “Codebase + Config Analysis” via scanning files) and to optionally read user-provided local .env values after explicit confirmation (Step 0.25), which constitutes outsider-authored free text ingestion at runtime.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill performs a blocking runtime MCP check and will call the Descope MCP server (https://mcp.descope.com) to fetch SDK method names and API signatures that directly influence generated prompts and code