diagnosing-bugs
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill facilitates the creation and execution of debugging scripts and regression tests. It specifically instructs the agent to use and modify the local
scripts/hitl-loop.template.shto drive human-in-the-loop reproduction, which involves script generation and execution within the local environment.\n- [INDIRECT_PROMPT_INJECTION]: The debugging process requires the ingestion of untrusted external data which could contain malicious instructions. The agent is guided to read logs, HAR files, and captured artifacts that may be attacker-controlled.\n - Ingestion points:
SKILL.mdinstructs the agent to read logs, network traces, and log dumps;scripts/hitl-loop.template.shcaptures error messages directly from the user's terminal output.\n - Boundary markers: Absent; the skill does not define delimiters or specific instructions to ignore embedded commands within the processed artifacts.\n
- Capability inventory: The skill enables shell execution (bash), CLI tool usage (curl, git), and the execution of various testing frameworks.\n
- Sanitization: The skill mandates secret redaction but does not specify validation or sanitization of the content of the diagnostic data to prevent prompt injection.\n- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill instructions involve capturing and displaying system artifacts and network traces. While there is a specific requirement to redact secrets using
<REDACTED>, the process inherently involves the handling and potential exposure of sensitive operational data during the diagnosis loop.
Audit Metadata