The Agent Skills Directory

[SAFE]: The skill instructions are focused on providing precise definitions from a local glossary. It strictly limits the agent's output to specific terms and definitions, which helps maintain predictable behavior and prevents deviation from its intended purpose.
[DATA_EXPOSURE]: The skill reads local reference files and instructs the agent to identify existing project component names. This is a functional requirement for providing context-aware naming and does not involve unauthorized access to sensitive system credentials or exfiltration of data.
[PROMPT_INJECTION]: The skill includes robust self-regulating instructions to prevent the agent from deviating from its role as a naming reference. It explicitly forbids providing recommendations, implementation details, or coining new terms, effectively creating narrow output boundaries.

naming