The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes several categories of repository-local documentation to generate its output and inform architectural decisions. This includes engine references, game design documents (GDDs), and existing ADR files.
Ingestion points: Reads files from docs/engine-reference/, design/gdd/, docs/architecture/, and .claude/docs/.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instruction' warnings when these files are interpolated into the LLM context.
Capability inventory: The skill has significant capabilities, including the ability to Write and Edit repository files, and the Task tool to spawn sub-agents for specialized reviews.
Sanitization: There is no evidence of content sanitization or validation for the text read from local documentation. However, the skill mitigates risk by requiring explicit user approval via AskUserQuestion before committing any changes to the ADRs or the architecture registry.

architecture-decision