The Agent Skills Directory

Indirect Prompt Injection: The skill ingests untrusted data from product teams and requirement documents to facilitate the assessment process. This pattern creates a surface for potential indirect prompt injection if the ingested data contains instructions designed to influence the agent's behavior. (1) Ingestion points: Product team intake questions and Product Requirement Documents (PRD) defined in the 'Intake' section of SKILL.md. (2) Boundary markers: No specific delimiters or 'ignore embedded instructions' warnings are mandated for the intake content. (3) Capability inventory: The skill can invoke research tools and perform file-write operations within matter-specific directories. (4) Sanitization: The skill relies on source attribution and attorney review gates rather than automated sanitization of external inputs.
File System Interaction: The skill reads and writes to specific paths within the ~/.claude/plugins/config/ directory to manage organizational templates and matter context. This behavior is consistent with the skill's operational requirements for maintaining project-specific documentation.

pia-generation