The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides the AI agent with specific shell commands using uv run to manage the BMad environment. These commands execute the skill's own Python scripts (setup.py) to handle installation, updates, and maintenance tasks as described in references/setup.md.
[EXTERNAL_DOWNLOADS]: The scripts/setup.py script includes functionality to check for updates by fetching manifest files from remote sources, including GitHub and other HTTPS URLs. By default, it targets the vendor's repository for version comparisons.
[INDIRECT_PROMPT_INJECTION]: The skill discovers and categorizes other installed skills by scanning their directories for module-manifest.toml files and reading host-provided descriptions. This represents an ingestion surface for data from other skills. The skill includes instructions to the agent to rely only on the provided references/help.md for routing logic and to limit the use of external metadata to mitigate potential influence from third-party skill content.
[DYNAMIC_EXECUTION]: The render_skill.py script automates the creation of project snapshots by rendering markdown templates. It processes configuration tokens and writes the resulting files to the project's runtime directory (_bmad/render).

bmad