The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts using the uv run command to manage its method catalog and resolve configuration settings. These scripts include scripts/pick_methods.py within the skill root and framework-level scripts like _bmad/scripts/resolve_customization.py and _bmad/scripts/resolve_config.py. These operations are scoped to local files and represent standard agent behavior for managing complex toolsets.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data, specifically conversation history and method definitions from customize.toml, as targets for its elicitation and refinement logic. This is the primary intended purpose of the skill and is handled via structured instructions that cast roles (personas) and apply reasoning patterns (e.g., Socratic questioning, Red Teaming) to the content. The implementation includes clear boundaries, such as halting for user approval before applying any suggested changes.
[EXTERNAL_DOWNLOADS]: The skill manifest references a GitHub repository for updates (github:bmad-code-org/bmad-skills/skills). This is a standard update mechanism for the bmad-code-org vendor and does not involve runtime execution of untrusted remote code.

bmad-advanced-elicitation