scientific-debugging
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external user data and system outputs which can contain untrusted instructions.
- Ingestion points: The agent is directed to collect user reports, error stacks (SKILL.md), and browser console errors (resources/browser-test-guide.md).
- Boundary markers: The instructions lack specific delimiters or commands to the agent to disregard instructions embedded within the processed bug reports or log files.
- Capability inventory: The skill uses tools for browser interaction (
browser_subagent) and code inspection (view_code_item), and it includes steps to apply code modifications. - Sanitization: There are no defined mechanisms for validating or sanitizing the content of user-provided reports or automated logs before ingestion.
Audit Metadata