The Agent Skills Directory

[SAFE]: The skill's logic is focused on professional audio orchestration and does not contain any malicious command execution, data exfiltration, or obfuscation. It includes explicit defensive instructions to exit if no arguments are provided and to seek user approval at every phase transition using the AskUserQuestion tool.
[INDIRECT_PROMPT_INJECTION]: The skill reads from various repository-local files to gather context for its sub-agents, which is a standard surface for orchestration tools. The potential risk of malicious instructions within those project files is mitigated by the structured review process.
Ingestion points: Reads repository instructions (AGENTS.md, CLAUDE.md, Assets/CLAUDE.md), design documents (design/gdd/), and project-specific configurations (production/review-mode.txt, .claude/docs/technical-preferences.md).
Boundary markers: The skill does not explicitly use delimiters or "ignore embedded instructions" warnings for the ingested data.
Capability inventory: The orchestration environment has access to Bash, Write, Edit, and Task (for sub-agent spawning).
Sanitization: No programmatic sanitization is present; however, the skill mandates user approval (AskUserQuestion) for all proposals generated from the ingested context.

team-audio