The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands including npx agenticloops, 5dive loop, and gh to manage the lifecycle of recurring agents. These commands are essential for the skill's primary function of scheduling and running jobs.
[EXTERNAL_DOWNLOADS]: The skill fetches the agenticloops CLI package via npx and downloads loop definitions from the agenticloops.dev directory and public GitHub repositories.
[REMOTE_CODE_EXECUTION]: The skill enables the installation and execution of external 'loops' which contain prompts and skill references. To mitigate risk, the instructions include a 'Vet before recommending' section that guides users to evaluate the loop's required permissions (CLI tools, secrets, network egress) and author reputation before scheduling them for unattended execution.
[SAFE]: The skill implements safe practices for sensitive data by ensuring that secrets are only referenced by environment variable names (e.g., X_API_TOKEN) within loop files, ensuring that actual credential values are never stored in the shared definitions. It also uses a 'declare-and-check' model for requirements to provide visibility into what a loop will access before it is run.

loops