unity-ai-audit

name	unity-ai-audit
description	Use to audit a Unity project's AI/ML stack: packages, models, scripts, configs, and release risks. Start here for "what AI does this game use?"

Unity AI stack audit

One-shot

Run get_ai_stack_summary with include_prompt_assets: true.

Deep dive order

Step	Tool	Why
1	get_ai_ml_package_inventory	Official Unity AI packages & versions
2	list_ml_model_assets	ONNX and other models in repo
3	list_ml_agents_training_configs	RL training YAMLs
4	find_ai_related_scripts	Code-level AI APIs
5	list_ai_prompt_or_config_assets	Prompt/RAG JSON or docs
6	get_build_size_estimate	Model weight in shipping build
7	get_release_readiness	Broken refs + cycles + large assets

Bundled skills

Call list_ai_skills (returns category) then read_ai_skill.

Skills follow Unity Editor domains — see skills/README.md. Highlights:

• Scenes: unity-scenes-workflow
• Model-agnostic: unity-model-agnostic-inference
• Vendor-specific: unity-ml-agents, unity-sentis-inference, unity-llm-integration
• Gameplay AI: unity-navmesh-npc
• Performance: unity-runtime-performance, unity-build-size-optimization, unity-memory-asset-pressure

Report template

## AI stack summary
- Packages: ...
- Models: N ONNX, ...
- Scripts: N files (top labels: ...)