The Agent Skills Directory

[SAFE]: The skill is primarily instructional, providing a four-phase framework for debugging technical issues. It does not include any malicious instructions or hidden payloads.
[COMMAND_EXECUTION]: The skill includes a bash script (find-polluter.sh) and markdown examples that execute local commands like npm test, find, security, and codesign. These are standard developer tools used for local diagnostic and build purposes and do not present a security risk in this context.
[DATA_EXFILTRATION]: Diagnostic examples for checking environment variables (e.g., ${IDENTITY:+SET}) use safe shell patterns to confirm the presence of secrets without printing their actual content. No hardcoded credentials or sensitive file path access (like SSH keys) were found.
[PROMPT_INJECTION]: While the skill contains strong behavioral directives for the AI agent (e.g., "The Iron Law"), these are intended to guide the agent's problem-solving process rather than bypass the AI's safety filters or instructions.

systematic-debugging