alibaba-supplier-outreach
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, specifically Alibaba supplier information and incoming messages, which could contain malicious instructions designed to manipulate the agent's behavior.
- Ingestion points: Supplier data is ingested via
mcp__launchfast__supplier_researchin the Outreach mode, and supplier replies are read from the browser usingmcp__claude-in-chrome__read_pagein the Check Replies and Negotiate modes. - Boundary markers: The instructions do not define boundary markers to separate supplier-provided text from the agent's own logic, which may lead the model to treat external content as instructions.
- Capability inventory: The agent has significant capabilities, including full browser automation via
mcp__claude-in-chrome__*(clicking, typing, navigation, screenshots) and the ability to write to the local file system at~/.claude/supplier-conversations/. - Sanitization: There is no logic provided to sanitize or filter the content retrieved from external pages before it is processed by the agent to determine negotiation stages or draft replies.
Audit Metadata