debugging-toolkit-smart-debug
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the $ARGUMENTS variable, which represents user-provided issue reports, stack traces, and logs. This data is used to drive the agent's analysis, hypothesis generation, and fix implementation.
- Ingestion points: Untrusted input from $ARGUMENTS is processed in the 'Context' section to extract error messages and reproduction steps, and again at the end of the instructions to define the 'Issue to debug'.
- Boundary markers: The skill lacks explicit delimiters (e.g., XML tags or triple quotes) or specific 'ignore embedded instructions' directives when interpolating the user input into the prompt context.
- Capability inventory: The skill utilizes a 'Task' tool with a 'debugger' subagent and provides instructions for the AI to perform code flow reconstruction, generate code fixes, and run test suites.
- Sanitization: No sanitization, escaping, or schema validation is applied to the incoming issue data before it is processed by the LLM.
Audit Metadata