The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface (Category 8) because it ingests untrusted user data to generate documentation intended for consumption by downstream automated skills in a multi-agent workflow.
Ingestion points: User descriptions and Q&A responses collected during the execution of instructions in SKILL.md.
Boundary markers: The output brief markdown template lacks delimiters or explicit instructions to isolate user content from the surrounding documentation structure.
Capability inventory: The skill is designed as the first step in a planning chain where its output files directly drive the behavior of subsequent skills such as create-prd and create-stories.
Sanitization: No data validation, filtering, or escaping mechanisms are implemented for the user-supplied content before it is written to the persistent docs/ storage.

create-brief