The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data, creating an attack surface where malicious content in project files could influence the agent's behavior.
Ingestion points: SKILL.md specifies reading project artifacts such as Game Design Documents (GDDs), Architecture Decision Records (ADRs), and consistency logs (docs/consistency-failures.md).
Boundary markers: The instructions do not define delimiters or specific warnings to the agent to ignore instructions embedded within these external files.
Capability inventory: The skill has access to Bash for running tests, Write for updating project stage files, and Task for spawning subagents.
Sanitization: The skill does not implement sanitization or validation of the content found in the project documentation before processing it as context.
[DYNAMIC_EXECUTION]: The skill dynamically spawns multiple subagents to evaluate project status.
Evidence: Section 4b of SKILL.md triggers the parallel execution of four director-level subagents (creative-director, technical-director, producer, and art-director) using the Task tool, with execution parameters and protocol derived from an external documentation file (.claude/docs/director-gates.md).

gate-check