matt-diagnosing-bugs

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted data including application logs, HAR files, network traces, and error reports.
  • Ingestion points: Data enters the agent's context through logs, traces, and manual user inputs provided during the debugging loop (SKILL.md).
  • Boundary markers: The instructions include a proactive requirement to redact secrets and only quote relevant signal lines, which helps limit the exposure of the full untrusted payload.
  • Capability inventory: The agent is authorized to execute arbitrary shell commands and specific diagnostic scripts (scripts/hitl-loop.template.sh) to facilitate debugging.
  • Sanitization: The skill mandates the use of '' placeholders for sensitive environment variables and credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 10:29 AM
Security Audit — agent-trust-hub — matt-diagnosing-bugs