The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands defined in the customize.toml configuration file through lifecycle hooks like activation_steps and external_handoffs. This mechanism allows for dynamic execution of shell commands based on project-specific configuration.- [COMMAND_EXECUTION]: During the activation phase, the skill executes a Python script located at {project-root}/_bmad/scripts/resolve_customization.py. This invokes external code stored within the project structure.- [DATA_EXFILTRATION]: The external_handoffs feature provides a capability to route data to external platforms like Confluence or Notion. While intended for documentation, this could be leveraged for unauthorized data transfer if the configuration is compromised or manipulated.- [PROMPT_INJECTION]: The skill is designed to read and interpret untrusted user inputs from documents such as game briefs and research notes. This ingestion of external data combined with the skill's capability to execute commands creates a surface for indirect prompt injection. Ingestion points: game-briefs, brainstorming-docs, research-notes. Boundary markers: Absent. Capability inventory: Shell command execution, Python script execution. Sanitization: Absent.

gds-gdd