The Agent Skills Directory

[SAFE]: The skill restricts its operations to local files within the ~/.gtm/ directory and contains explicit instructions to avoid network access and external system changes.
[SAFE]: Data safety is prioritized via instructions in references/context.md that forbid the echoing or storage of sensitive items like API keys or session tokens found in processed data.
[SAFE]: The execution flow is strictly controlled, resolving context and persona visibility before performing classification without the use of dangerous tools or remote execution capabilities.

gtm-lead-segmentation