The Agent Skills Directory

[SAFE]: The skill contains purely instructional content for diagnosing software defects. It provides a classification table for Roblox symptoms and a logical loop for debugging (observation, environment, hypothesis, reproduction).
[SAFE]: No external network operations, credential exposures, or dangerous command execution patterns were found. The skill operates by guiding the agent through reading existing code and performing playtests within a development environment.
[SAFE]: All external references use local relative paths (e.g., ../best-practices/references/) and point to shared documentation within the project structure, representing standard documentation practices.
[SAFE]: The skill includes explicit security and alignment guardrails, such as Rule 11 ('User authority outranks this skill') and the 'User Authority' section, which strictly forbid the agent from making autonomous code changes or removing user-placed probes.

diagnose