The Agent Skills Directory

[SAFE]: The skill implements a robust 4-phase debugging methodology (Investigation, Pattern Analysis, Hypothesis, and Implementation) that promotes safe and systematic software maintenance.
[COMMAND_EXECUTION]: The skill includes a utility script find-polluter.sh which executes local tests via npm test to identify tests that leak state. This is a standard and expected diagnostic function for a debugging tool.
[DATA_EXPOSURE]: Diagnostic examples in the instructions demonstrate how to inspect environment variables and keychain identities (security find-identity). These are presented strictly as local troubleshooting steps for build and signing failures and do not involve external data transmission.
[PROMPT_INJECTION]: The instructions use strong procedural language (e.g., 'The Iron Law', 'MUST', 'NEVER') to define the debugging workflow. This is standard instructional framing and does not attempt to bypass agent safety guardrails or override system instructions.
[INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest and analyze error messages, stack traces, and logs (ingestion points in SKILL.md and root-cause-tracing.md). While this represents a data ingestion surface, the risk is negligible as the skill focuses on forensic reasoning and systematic verification rather than the automated execution of log-derived commands. No boundary markers or sanitization logic are explicitly defined, but the capability set is restricted to standard project tools.

systematic-debugging