The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill is instructed to execute arbitrary steps defined in the activation_steps_prepend and activation_steps_append arrays. These steps are loaded dynamically from configuration files (customize.toml and its overrides) at runtime.
Evidence: SKILL.md Step 2 and Step 7 explicitly command the agent to execute each entry in these arrays in order.
[COMMAND_EXECUTION]: The skill executes Python scripts located within the project's directory structure using the uv run command.
Evidence: SKILL.md Step 1 and Step 5 run {project-root}/_bmad/scripts/resolve_customization.py and {project-root}/_bmad/scripts/resolve_config.py respectively.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external configuration files and project-level files, which are then used as foundational facts or executable instructions without sanitization.
Ingestion points: customize.toml, {project-root}/_bmad/custom/{skill-name}.toml, and various files referenced via globs in the persistent_facts array.
Boundary markers: None. The instructions do not include delimiters or warnings to ignore instructions embedded within the ingested facts or configuration files.
Capability inventory: The skill possesses the ability to execute shell commands (via uv run), read project files, and perform sequence-based activation steps.
Sanitization: No validation or filtering is applied to the content of the configuration files or the persistent_facts before they are processed by the agent.

bmad-agent-architect