The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from various repository-local files to guide the UI design and implementation pipeline.
Ingestion points: The skill reads production/review-mode.txt, design/gdd/game-concept.md, design/player-journey.md, design/ux/interaction-patterns.md, and design/accessibility-requirements.md in references/claude-skill.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between data and potentially malicious instructions within these files.
Capability inventory: The skill is permitted to use Bash, Write, Edit, Task, and TodoWrite tools, providing a significant capability surface for actions triggered by processed data.
Sanitization: The skill does not implement sanitization or filtering for the content read from the external documentation files before interpolation into the agent's context.
[COMMAND_EXECUTION]: The skill explicitly allows the use of the Bash tool and the Task tool for delegating work to subagents. While these are part of the intended orchestration functionality, they represent a powerful execution environment that could be targeted by the aforementioned indirect prompt injection vector.

team-ui