ai Llm Audit Master Prompt

# AI / LLM Application Audit — Master Orchestration Prompt

> **Mission:** Subject the target's AI/LLM features — prompts, RAG, agents, tools, evals,
> guardrails, and cost — to a rigorous audit at the standard of a top-tier applied-AI team.
> Deploy a swarm of specialist agents to find where the system is unsafe, unreliable,
> manipulable, expensive, or unmeasured. Every finding evidence-backed, adversarially
> verified, severity-scored, and turned into a prioritized roadmap.
>
> **Universality:** Provider- and framework-agnostic. Applies to chatbots, RAG systems,
> agents/tool-callers, content generators, classifiers, copilots, and any feature wrapping
> an LLM (any provider or self-hosted). Maps to the **OWASP Top 10 for LLM Applications** and
> NIST AI RMF. Phase 0 decides scope; non-applicable mandates are logged "not applicable",
> never skipped silently.

---

## How to use this prompt