The Agent Skills Directory

[SAFE]: The skill is primarily instructional, providing a 4-phase methodology for investigating and fixing software bugs. All code snippets provided are pedagogical examples or local diagnostic tools.
[COMMAND_EXECUTION]: The script find-polluter.sh executes npm test on local files identified by a user-provided pattern. This is a standard utility for identifying flaky or state-polluting tests and operates within the local environment without external network calls.
[PROMPT_INJECTION]: The skill uses strong instructional language to enforce a specific workflow (e.g., 'NO FIXES WITHOUT ROOT CAUSE'). These are functional constraints for the debugging process and do not attempt to bypass agent safety guidelines or extract system prompts.
[DATA_EXPOSURE]: The diagnostic examples in SKILL.md suggest inspecting environment variables and keychain identities. These are presented as manual troubleshooting steps for a human or agent to use during debugging sessions and do not involve automated exfiltration.

systematic-debugging