The Agent Skills Directory

[COMMAND_EXECUTION]: The skill references standard diagnostic and development commands, such as npm, bun, security, codesign, and env, as examples of how to gather evidence during the debugging process. These are used in a transparent and contextually appropriate manner for the described debugging scenarios.\n- [PROMPT_INJECTION]: The skill instructs the agent to process potentially untrusted external data, including error messages, stack traces, and version control diffs. There are no specific boundary markers or sanitization steps defined for these inputs. However, this represents a standard operational surface for a debugging agent and no malicious intent was detected.\n- [SAFE]: The skill implements a constructive debugging methodology without any detected malicious patterns, obfuscation, or unauthorized data access. The provided commands are standard for development environments and are intended for legitimate diagnostic use.

systematic-debugging