ai-on-saas-procurement-and-questionnaire

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

• Third-party content exposure detected (high risk: 0.85). Required Inputs include “the buyer's questionnaire (if issued)” (SKILL.md:21), i.e., outsider-authored free-form procurement questionnaire text is read and then used to generate the response pack (workflow steps 1–2 at SKILL.md:28-30), creating an indirect prompt-injection exposure path.

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

• Potentially malicious external URL detected (high risk: 0.90). The file explicitly instructs "Load ../ai-agent-procurement-and-questionnaire/SKILL.md" (../ai-agent-procurement-and-questionnaire/SKILL.md), indicating a runtime dependency that would be fetched to extend the agent's answer pack and thus directly influence prompts/instructions.