systematic-debugging
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a utility script
find-polluter.shwhich executes shell commands (npm test) to identify problematic test files. This is standard functionality for a debugging toolkit. - [INDIRECT_PROMPT_INJECTION]: The debugging methodology requires the agent to ingest and analyze external data such as error logs and stack traces. This creates a surface where malicious instructions embedded in logs could potentially influence the agent, although this is an inherent risk of the debugging task.
- [INDIRECT_PROMPT_INJECTION]: Ingestion points: Error logs, stack traces, and build outputs (referenced in
SKILL.mdandroot-cause-tracing.md). - [INDIRECT_PROMPT_INJECTION]: Boundary markers: None specified.
- [INDIRECT_PROMPT_INJECTION]: Capability inventory: Shell script execution via
find-polluter.shand diagnostic instrumentation. - [INDIRECT_PROMPT_INJECTION]: Sanitization: None specified.
Audit Metadata