diagnosing-bugs
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as network traces, HAR files, and log dumps to diagnose bugs, creating an ingestion surface for indirect instructions.\n
- Ingestion points: Captured traces, log dumps, and HAR files processed in Phase 1 and the troubleshooting sections of
SKILL.md.\n - Boundary markers: The skill does not define specific delimiters for ingested logs, although it instructs the agent to focus only on signal-bearing lines.\n
- Capability inventory: The agent is instructed to execute shell commands, run curl scripts, and perform file operations to build feedback loops.\n
- Sanitization: The skill includes a dedicated 'Redact' section that mandates replacing secrets with
<REDACTED>before showing output to prevent credential leakage.\n- [DYNAMIC_EXECUTION]: The skill's primary function involves generating and running dynamic code to reproduce and fix bugs.\n - Evidence: Phase 1 of
SKILL.mdexplicitly instructs the agent to build throwaway harnesses, HTTP scripts, and CLI invocations to test hypotheses.\n - Evidence: The skill provides a bash template (
scripts/hitl-loop.template.sh) designed to be customized and executed at runtime to facilitate human-in-the-loop debugging.
Audit Metadata