The Agent Skills Directory

[SAFE]: The skill performs read-only operations on local workspace files, such as .claude/state/ and config/ai-tools/, to summarize internal state for the user.
[SAFE]: No network activity, external data transmissions, or remote code execution patterns are present.
[SAFE]: The prompt instructions are focused on workspace management and do not attempt to bypass safety filters or override system constraints.
[SAFE]: The instructions for path discipline and session integration are standard developer workflows and do not pose a threat to the host system.

session-start