alibaba-supplier-outreach

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface exists where the skill ingests supplier messages. Ingestion points: Message threads are read from the Alibaba UI during the 'REPLY REVIEW' phase. Boundary markers: Absent; there are no specific instructions to the agent to ignore or delimit instructions found within supplier replies. Capability inventory: The skill utilizes browser automation for navigating, clicking, and submitting forms, and performs file-write operations to the './artifacts/' directory. Sanitization: Absent; content extracted from messages is summarized and processed directly. Mitigation: The risk is significantly lowered as the skill requires explicit human approval before any message is sent through the browser.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 01:59 PM
Security Audit — agent-trust-hub — alibaba-supplier-outreach