The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a utility script find-polluter.sh which executes npm test on files discovered via bisection to identify environment-polluting tests. Additionally, SKILL.md contains examples of diagnostic commands for macOS environments, such as security list-keychains, security find-identity, and codesign, which are used to debug code-signing issues at the workflow level.
[SAFE]: The skill is primarily instructional, outlining a four-phase debugging process (Root Cause Investigation, Pattern Analysis, Hypothesis, and Implementation). The included 'pressure tests' (test-pressure-1.md through test-pressure-3.md) are behavioral evaluation scenarios designed to verify the agent's adherence to the skill's logic under simulation, rather than malicious prompt injections aimed at bypassing core system safety.
[SAFE]: Data validation patterns in defense-in-depth.md and polling patterns in condition-based-waiting.md promote standard software engineering best practices for stability and security.

systematic-debugging