browse

Warn

Audited by Snyk on Feb 25, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 1.00). The skill docs (e.g., skills/browse/SKILL.md and skills/functions/SKILL.md) explicitly instruct the agent to open arbitrary external URLs (browse open / browse --ws $BROWSERBASE_CONNECT_URL open ) and include code examples that call page.goto(params.url), snapshot, and page.evaluate()—including scraping public sites like news.ycombinator.com—so untrusted, user‑generated third‑party content is fetched, interpreted, and can directly influence subsequent actions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 25, 2026, 06:07 PM