The Agent Skills Directory

[SAFE]: The skill is a comprehensive developer manual for the 'lookbook' package. It defines architecture layers and extension points for scorers, filters, and selectors without introducing any malicious instructions or bypasses.
[PROMPT_INJECTION]: The skill presents a potential surface for indirect prompt injection because it instructs the agent to ingest data from local project files (e.g., misc/docs/lookbook_development_plan.md) and image metadata. This is an inherent part of a developer-focused skill and does not represent an active threat.
Ingestion points: Documentation files in the misc/docs/ directory and image metadata provided through the ImageRef protocol.
Boundary markers: None specified in the instructions for delimiting untrusted content.
Capability inventory: The skill describes capabilities for file mapping via dol, registry-based class instantiation, and network service exposure via a built-in HTTP server.
Sanitization: No specific sanitization or validation logic for external content is described.
[COMMAND_EXECUTION]: Technical references to CLI tools (e.g., lookbook serve, lookbook mcp) are provided for standard development and operational tasks within the package ecosystem.

lookbook-dev