systematic-debugging
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides
find-polluter.sh, a utility script for identifying state-polluting tests by looping through a directory and executingnpm test. The instructions also include examples of diagnostic shell commands for verifying environment variables and code-signing identities. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data sources like error logs and system output during the debugging phase.
- Ingestion points: Error messages, stack traces, and component logs in
SKILL.md. - Boundary markers: Absent; the skill relies on the agent's ability to interpret technical output.
- Capability inventory: Writing and running reproduction scripts/tests, and executing diagnostic commands.
- Sanitization: Not explicitly performed on analyzed logs.
Audit Metadata