Skills
skills/practicalswan/agent-skills/systematic-debugging/Gen Agent Trust Hub

systematic-debugging

Pass

Audited by Gen Agent Trust Hub on Feb 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The shell script find-polluter.sh automates the identification of problematic tests by executing npm test on files discovered via a user-specified search pattern. This utility is intended for local development and debugging of test suites.
  • [PROMPT_INJECTION]: The skill includes evaluation files (test-pressure-1.md, test-pressure-2.md, and test-pressure-3.md) that use role-playing and urgency-based prompts to test whether the agent will deviate from its core debugging rules. While these utilize injection-style framing (e.g., 'IMPORTANT: This is a real scenario'), they are clearly marked as internal validation tools for the skill's own logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 26, 2026, 10:13 PM