The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses authoritative and prescriptive language, referred to as the 'Iron Law,' to mandate a specific operational procedure for debugging. These instructions are designed to prevent the AI from taking shortcuts or guessing fixes, thereby improving the reliability of its technical assistance. They do not attempt to bypass system safety filters or jailbreak the underlying model.
[COMMAND_EXECUTION]: The skill provides a bash script named find-polluter.sh which identifies tests creating unwanted state by executing npm test on local files. Additionally, the documentation includes example commands for debugging build environments, such as codesign and security on macOS. These are standard development utilities appropriate for the skill's primary objective.
[DATA_EXFILTRATION]: In the context of debugging multi-component systems, the skill demonstrates a safe method for verifying the presence of environment variables (e.g., ${IDENTITY:+SET}${IDENTITY:-UNSET}) without outputting their actual values, which follows security best practices for secret management during diagnostics.

systematic-debugging