The Agent Skills Directory

[SAFE]: The skill is purely instructional and serves as a routing layer for other tools within the CodeStable ecosystem. It focuses on identifying user needs and directing them to appropriate sub-skills.
[COMMAND_EXECUTION]: The skill instructions specify using Glob and Read tools to inspect the local codestable/ directory structure and documentation. These operations are limited to project-specific metadata and do not involve arbitrary shell command execution or system-level modifications.
[DATA_EXFILTRATION]: No network-enabled tools or operations are requested or utilized. The skill operates entirely on local context and provides verbal recommendations to the user.
[PROMPT_INJECTION]: The instructions are clear and focused on logical routing. There are no attempts to bypass safety filters, extract system prompts, or override agent constraints.

cs