The Agent Skills Directory

[SAFE]: The skill focuses on improving developer workflows by establishing a four-stage debugging process (Investigation, Pattern Analysis, Hypothesis, and Implementation). It contains no remote connections, obfuscated code, or unauthorized data access mechanisms.
[COMMAND_EXECUTION]: The skill includes a shell script, find-polluter.sh, which automates the execution of local test files using npm test. This is a standard utility for local development and debugging of test suites, with no evidence of command injection or malicious payload delivery.
[DATA_EXFILTRATION]: Documentation in SKILL.md provides examples of how to verify environment variables and system identities (e.g., security list-keychains) during troubleshooting. These are presented as manual developer diagnostics for fixing build or environment issues and do not involve automated exfiltration to external targets.
[PROMPT_INJECTION]: The instructions use strong, prescriptive language to discourage the AI agent from proposing quick, non-investigated fixes. While restrictive of the agent's behavior, this steering is entirely aligned with the skill's purpose of preventing 'symptom-fixing' and ensuring technical rigor.

systematic-debugging