cloud-agent

Warn

Audited by Snyk on Mar 6, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.90). The SKILL.md deep-research workflow explicitly states the cloud agent will "爬取多个竞品网站" (crawl multiple competitor websites) as part of its task, meaning it fetches and ingests untrusted public web content which the agent will read and use to drive analysis and actions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 6, 2026, 04:19 PM