The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a Python script during activation using uv run at {project-root}/_bmad/scripts/resolve_customization.py. This is a vendor-provided mechanism for merging configuration layers. \n- [INDIRECT_PROMPT_INJECTION]: The skill has a significant indirect injection surface. \n
Ingestion Points: update-standards.md accepts user preferences, write-document.md performs web searches and research, and validate-doc.md reviews external documentation. \n
Boundary Markers: The instructions do not define delimiters or specific 'ignore instructions' warnings for processing these external inputs. \n
Capability Inventory: The agent has the ability to write to files (documentation-standards.md), execute framework commands (uv run), and invoke subprocesses for research. \n
Sanitization: No sanitization or content validation logic is present in the instructions. \n- [INDIRECT_PROMPT_INJECTION]: The sub-skill update-standards.md allows user-provided rules to be saved into documentation-standards.md as 'User Specified CRITICAL Rules'. These rules are subsequently treated as high-priority instructions for other tasks, creating a vector for persistent instruction injection. \n- [DYNAMIC_EXECUTION]: The skill utilizes subprocesses for web searching, research, and document validation. Additionally, the activation flow executes arbitrary steps defined in the activation_steps_prepend and activation_steps_append configuration blocks.

gds-agent-tech-writer