---
title: 'Mitigations/ web articles'
url: 'https://grav.munn.me/ai/mitigations'
markdown: 'https://grav.munn.me/ai/mitigations.md'
date: '2025-11-09'
description: 'This page describes mitigations and AI model safety. Topic Description Rating Dual LLM prompt filterer This explores the topic of using 2 models 1 supervisory and privileged and 1 non-privileged for prompt safety managemnt Centralized MCP governeror Lets you do all your MCP enforcement fr…'
---

# Mitigations/ web articles

This page describes mitigations and AI model safety.

| Topic | Description | Rating |
|---|---|---|
| [Dual LLM prompt filterer](https://simonwillison.net/2025/Apr/11/camel/) | This explores the topic of using 2 models 1 supervisory and privileged and 1 non-privileged for prompt safety managemnt |  |
| [Centralized MCP governeror](https://github.com/grafbase/nexus) | Lets you do all your MCP enforcement from a single place. |  |
| [LLM code reviwer](https://github.com/llimllib/personal_code/blob/c85d6b2106e1369339d045e8be0e43484149848e/homedir/.local/bin/review) | lets you use llm to review changes in code |  |
| [Agents rule of Two Paper](https://simonwillison.net/2025/Nov/2/new-prompt-injection-papers/#atom-everything) | rule of two to help with the lethal trifecta |  |
| [SELF-SUPERVISED INFERENCE OF AGENTS IN TRUSTLESS ENVIRONMENTS](https://arxiv.org/pdf/2409.08386v1) | good paper to read on self-supervised agents |  |
| [why models hallucinate](https://arxiv.org/pdf/2509.04664) | good paper on why ai models are somewhat untrustworthy |  |

---

## Navigation

- Parent: [AI](https://grav.munn.me/ai.md)
- Next: [Useful AI Prompts](https://grav.munn.me/ai/useful-ai-prompts.md)