The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs strong instructional language (e.g., "The Iron Law," "Violating the letter of this process is violating the spirit") to mandate a specific debugging methodology. These are legitimate behavioral constraints for technical task performance and do not attempt to bypass agent safety filters or override system-level instructions.
[DATA_EXFILTRATION]: The skill provides diagnostic examples for debugging build and signing processes. One example uses env | grep IDENTITY to verify environment variable propagation. While this could potentially log a secret in the session output if the variable contains one, it is a standard debugging technique for code-signing issues and does not involve exfiltration to external domains.
[COMMAND_EXECUTION]: The inclusion of find-polluter.sh allows the agent to find tests that create persistent side effects. The script uses standard utilities (find, sort, npm test) and implements quoting for variable expansion, minimizing the risk of accidental command injection during path resolution.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data like error messages, stack traces, and source code. This represents a potential surface for indirect injection; however, the skill's methodology of "Systematic Debugging" (Phase 1-4) encourages analytical processing of data rather than direct obedience to content, which mitigates the risk.
[OBFUSCATION]: Analysis of all markdown and script files revealed no signs of Base64 encoding, zero-width characters, homoglyphs, or other obfuscation techniques meant to hide malicious intent or URLs.

systematic-debugging