The Agent Skills Directory

[SAFE]: The skill operates entirely within the local project environment, specifically targeting the .pawbytes directory for configuration, memory, and artifact storage. It uses a collection of local reference files (e.g., offer-packaging.md, delivery-design.md) to guide its logic, ensuring that its behavior is predictable and constrained to the project scope.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to ingest data from user-controlled files to inform its service design process. This behavior is central to its primary function.
Ingestion points: Data is read from curated/product-context.md and curated/audience-intelligence.md within the paw-ps-sidecar memory directory, as detailed in SKILL.md.
Boundary markers: The skill does not explicitly define or use boundary markers (like XML tags or specific delimiters) to isolate ingested text from its core instructions.
Capability inventory: The skill's capabilities are limited to reading and writing markdown files within the resolved project paths, as described in the "Path Resolution" section of SKILL.md and autonomous-execution.md.
Sanitization: No specific sanitization or filtering logic is applied to the ingested content before it is processed by the agent.

paw-ps-service-executor