The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from project source code and user interactions to maintain documentation files.\n- Ingestion points: The skill reads project source code files to ensure the domain model matches the implementation and takes input from user dialogue for terminology (SKILL.md).\n- Boundary markers: There are no specific delimiters or instructions to ignore potential commands embedded within the code or comments that the agent is instructed to read.\n- Capability inventory: The skill has the capability to write to the local file system to update ZJ-CONTEXT.md and create Architectural Decision Records (ADRs) in the docs/zj-adr/ directory.\n- Sanitization: The instructions do not define any sanitization, escaping, or validation procedures for content extracted from source code or user input before it is written into the documentation.

zj-domain-modeling