browser-subagent-task-bounding
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of markdown documentation and templates for optimizing browser-based agent tasks. It does not include any executable scripts, network operations, or sensitive data access.
- [INDIRECT_PROMPT_INJECTION]: The skill addresses the risks associated with autonomous browser agents ingesting untrusted web content. It provides a protocol to limit the scope of data ingestion (using specific DOM selectors) and execution (step ceilings), which serves as a security best practice to mitigate resource exhaustion and accidental obedience to instructions found on web pages.
Audit Metadata