The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The find-polluter.sh script uses npm test to isolate specific test files that may be creating unwanted state. This is a legitimate development utility designed for local environment cleanup and debugging.
[DATA_EXFILTRATION] (SAFE): Diagnostic examples in SKILL.md show how to verify the existence of environment variables and list macOS signing identities. These operations are restricted to the local context for build troubleshooting and do not involve any network calls to exfiltrate sensitive data.
[PROMPT_INJECTION] (SAFE): The skill utilizes strong instructional language (referred to as the 'Iron Law') to mandate a specific debugging process. These instructions are functional and intended to prevent agent shortcuts, rather than attempting to bypass safety filters or ignore system constraints.
[REMOTE_CODE_EXECUTION] (SAFE): The skill does not contain any patterns involving the download or execution of remote code. All scripts and logic are self-contained within the provided files.

systematic-debugging