The Agent Skills Directory

[COMMAND_EXECUTION]: During activation, the skill executes a customization resolver script (resolve_customization.py) located in the project's scripts directory using the uv run command.
[COMMAND_EXECUTION]: The skill framework supports "activation steps" (activation_steps_prepend and activation_steps_append) which allow for executing sequences of commands during the agent initialization process.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill includes a mechanism to load persistent facts from local files in the project directory using a file: prefix and glob patterns, which are then incorporated into the agent's context.

bmad-cis-agent-storyteller