The Agent Skills Directory

[DATA_EXPOSURE]: The skill reads game design documents from the design/gdd/ directory and session state from production/. These operations are limited to the local project workspace and are necessary for the skill's intended purpose of document analysis and generation.
[COMMAND_EXECUTION]: The skill utilizes the Task tool to trigger validation gates by spawning sub-agents (e.g., technical-director, producer, creative-director). These are controlled, internal orchestration calls defined within the skill's documented workflow.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests and processes content from design/gdd/game-concept.md, which is user-controlled data.
Ingestion points: Reads design/gdd/game-concept.md and other GDD files.
Boundary markers: None identified in the prompt templates for separating external data from instructions.
Capability inventory: The skill has Write, Edit, and Task capabilities, allowing it to modify files or invoke other agent processes.
Sanitization: No explicit sanitization or instruction-filtering of the ingested documentation is performed.

map-systems