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 executing npm 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
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:03 PM
Security Audit — agent-trust-hub — systematic-debugging