debugger
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for systematic troubleshooting and evidence gathering. It mandates a 'Never Fix' policy and requires the removal of all debugging artifacts (e.g., console logs, temporary files) after investigation, which aligns with security best practices for diagnostic agents.- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface due to its core function of analyzing external data sources.
- Ingestion points: Processes external data from Docker logs, application log files, and web pages via Playwright (SKILL.md).
- Boundary markers: The instructions lack explicit delimiters or instructions to ignore embedded prompts within the logs or web content being analyzed.
- Capability inventory: The agent can execute shell commands via Bash and interact with web browsers via Playwright (SKILL.md).
- Sanitization: There is no mention of sanitizing or filtering external inputs before processing or reporting.- [COMMAND_EXECUTION]: The skill uses Bash for diagnostic purposes (e.g., docker compose logs, grep, tail, curl). These commands are standard for a debugging role and do not involve privilege escalation or unauthorized persistence attempts.
Audit Metadata