course-guide

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow ingests user free text as the “ask” to interpret (Step 1, “Interpret the ask, which arrives…”), and then it may fetch course content docs/README based on that input, creating an indirect prompt-injection exposure surface.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to fetch the course README at https://raw.githubusercontent.com/rohitg00/ai-engineering-from-scratch/main/README.md at runtime (if local copy is absent) and uses that fetched content as the single source of truth to route and generate responses, so the remote file directly controls agent output.