The Agent Skills Directory

[SAFE]: The skill defines a standard methodology for requirement engineering, including traceability, lifecycle management, and human oversight checkpoints.
[INDIRECT_PROMPT_INJECTION]: The skill processes high-level product intent and discovery artifacts provided by the user to generate structured requirements.
Ingestion points: User product intent, feature concepts, and discovery artifacts (OBS/INS/HYP/EXP).
Boundary markers: Not explicitly defined for user inputs, but outputs are formatted in Markdown and YAML.
Capability inventory: File writing to the local project directory (e.g., .agile-v/REQUIREMENTS.md). No network or command execution capabilities are utilized.
Sanitization: Standard LLM processing of text to structured documentation format.
[DATA_EXFILTRATION]: The skill writes architectural blueprints and logs to the project-local .agile-v/ directory. This is used for project state management and does not involve accessing sensitive system credentials or exfiltrating data via network.

requirement-architect