The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requests permission for various bash tool namespaces including git, beans, gh, linear, npx, and mkdir. These are intended to support the integration of the task backends (GitHub, Linear, etc.) and file system management for the tool's local configuration.
[DATA_EXPOSURE]: The skill reads and writes configuration files within the local .groove/ directory. It also reads the version from a local skills/groove/SKILL.md file. No sensitive system files, environment variables, or hardcoded credentials are accessed or exposed.
[PROMPT_INJECTION]: The skill uses AskUserQuestion to gather configuration values. While it processes user input, the logic is constrained to specific keys and valid options defined in the instructions, and it does not allow for bypassing its own internal logic or the underlying agent's safety guidelines.

groove-admin-config