The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses backticks and bash blocks in the Preamble and post-execution sections to run local binaries such as rmbc-update-check, rmbc-analytics, and rmbc-workspace. It dynamically resolves the root path for these tools based on environment variables and standard installation paths.
[EXTERNAL_DOWNLOADS]: The skill includes an 'Inline upgrade flow' and update checks that reference external resources and scripts managed within the $_RMBC_ROOT directory. It also provides a command to open a YouTube URL via the open command.
[DATA_EXFILTRATION]: The skill contains a backgrounded analytics command (rmbc-analytics log) that sends metadata about skill usage, product names, and tiers. While there is a telemetry opt-in/opt-out prompt, the initial preamble attempts to run this analytics command if analytics_enabled is set to true (default) in the local config.yaml.
[PERSISTENCE_MECHANISMS]: The skill creates a hidden directory ~/.rmbc-skills and various state files (.intro-seen, .telemetry-prompted) to track user interactions and preferences across sessions. It also uses sed to modify ~/.rmbc-skills/config.yaml to toggle features like analytics.
[DYNAMIC_CONTEXT_INJECTION]: The Preamble uses shell command execution to determine the environment and state before the AI processes the prompt, replacing variables and echoing status flags directly into the LLM context.

webinar-registration-copy