The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses uv run to execute a Python script (resolve_customization.py) located within the project's _bmad directory to handle configuration merging. Additionally, the instructions encourage the use of subprocesses and subagents for artifact analysis.
[PROMPT_INJECTION]: The checklist.md file employs instructional override patterns, commanding the agent to adopt a "FRESH CONTEXT" as an "independent quality validator." It uses high-pressure language ("CRITICAL MISSION," "PREVENT DISASTERS") to override the primary instruction set for the validation phase.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process a wide range of external project files, including Game Design Documents, Architecture files, and Epics.
Ingestion points: Files such as sprint-status.yaml, epics.md, gdd.md, architecture.md, and project-context.md are loaded into the agent context.
Boundary markers: The instructions do not provide explicit delimiters or "ignore" directives to prevent the agent from obeying instructions potentially embedded within these documents.
Capability inventory: The skill possesses file read/write capabilities, shell command execution via uv, and network access through "web research" tasks.
Sanitization: There is no evidence of sanitization or content validation for data retrieved from project artifacts before it is used in prompt generation or output creation.
[DYNAMIC_EXECUTION]: The workflow supports a on_complete hook in the customize.toml configuration, which allows for the dynamic resolution and execution of shell commands at the end of the story creation process.

gds-create-story