The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the uv utility to execute local configuration and state management scripts, such as resolve_customization.py and memlog.py, located within the project's internal _bmad directory.
[DYNAMIC_EXECUTION]: The scripts/resolve_personas.py script uses the subprocess.run module to programmatically invoke other internal Python components to resolve agent and persona information. This is performed using the current system executable and strict argument lists.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external project files and materials to verify user claims, which presents a potential surface for indirect prompt injection.
Ingestion points: Project files and relevant materials are read into the agent context in SKILL.md to verify user-provided summaries.
Boundary markers: No explicit delimiters or specialized warnings for handling the content of these external files are specified in the instructions.
Capability inventory: The skill uses uv run to execute scripts that can write and append to the local file system (e.g., memlog.py and creating forged-idea.md).
Sanitization: The skill does not define specific sanitization or filtering logic for the ingested file content, relying instead on its skeptical questioning posture.

bmad-forge-idea