backlinks-summary

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

• Third-party content exposure detected (high risk: 0.85). The required workflow calls a runtime HTTP endpoint and then reads the returned response body (curl ... -o "$RESPONSE_FILE" followed by raw = response_path.read_text(...) and cat "$RESULT_FILE"), so outsider-authored free text (the provider’s JSON response) is ingested into the agent’s LLM context via stdout.