The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local Python scripts (resolve_customization.py and resolve_config.py) using uv run during its activation process. These scripts are located within the project's internal directory structure ({project-root}/_bmad/scripts/).
[DYNAMIC_EXECUTION]: The skill is configured to execute arbitrary lists of instructions defined in the activation_steps_prepend and activation_steps_append fields within its configuration files. This allows the agent's behavior to be dynamically extended at runtime based on the project's settings.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external configuration files and project files to establish its persona and foundational knowledge, which could be exploited if those files contain malicious instructions.
Ingestion points: Instructions and facts are loaded from customize.toml, {skill-name}.toml, and {skill-name}.user.toml, as well as file contents specified in the persistent_facts array.
Boundary markers: None; instructions in SKILL.md direct the agent to adopt the loaded persona and facts as foundational context without explicit delimiters or validation.
Capability inventory: The skill has the capability to execute shell commands (via uv run) and read files from the project directory.
Sanitization: There is no evidence of sanitization or filtering applied to the instructions or file contents before they are processed by the agent.

bmad-agent-architect