The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script, resolve_customization.py, using uv run during both the activation and completion phases. This script is part of the vendor's infrastructure located in the project's _bmad directory.\n- [DYNAMIC_EXECUTION]: Command execution targets computed paths within the project structure ({project-root}/_bmad/scripts/), which is a standard pattern for this vendor's automation tools.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external configuration files and allows arbitrary project files to be loaded into the agent's context as persistent facts.\n
Ingestion points: Configuration is merged from customize.toml, {skill-name}.toml, and config.yaml. The persistent_facts feature allows loading the contents of any file specified via the file: prefix.\n
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the ingested configuration or fact data as potentially untrusted.\n
Capability inventory: The skill is capable of directory and file creation, as well as executing local commands via uv.\n
Sanitization: There is no evidence of validation or sanitization for the external data loaded into the prompt context.

gds-test-framework