The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from project files, creating an attack surface where malicious instructions embedded in those files could influence agent behavior.
Ingestion points: Reads content from design/gdd/game-concept.md, design/game-brief.md, design/gdd/game-pillars.md, and design/gdd/systems-index.md during Phase 1.
Boundary markers: No specific delimiters or "ignore instructions" wrappers are defined for processing these documents.
Capability inventory: The skill has broad capabilities including file system modification (Write, Edit), shell execution (Bash), and spawning sub-agents (Agent).
Sanitization: No explicit sanitization or validation of the ingested markdown content is performed before processing.
[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes dynamic context injection to execute a shell command at load time, which replaces the placeholder with the command's output.
Evidence: !bash "${CLAUDE_SKILL_DIR}/../../hooks/yaml-helper.sh" resolve_config --keys review_mode,automation,workflow,docs.density,project.stage in SKILL.md.
Context: This command targets a local hook script to resolve project-specific configuration. While this pattern can be risky, its application here appears benign and workflow-oriented.

map-systems