The Agent Skills Directory

[SAFE]: The skill manages project context locally by storing data in the user's home directory under ~/.claude/ck/. It does not perform any network operations or access sensitive system files beyond standard project metadata files like package.json or .git/config.
[COMMAND_EXECUTION]: The skill executes git commands via spawnSync to retrieve repository information and commit counts. These commands are executed safely using argument arrays rather than shell strings, which prevents command injection.
[PROMPT_INJECTION]: The skill processes project-specific files like CLAUDE.md and README.md to build an automated context briefing. While these files are external data sources, the skill treats them as structured input for context tracking and presents data to the user for confirmation during initialization, which is standard behavior for project management tools.

ck