From ec7337feb50ac9052f478a09e42825df4a0dbcb5 Mon Sep 17 00:00:00 2001 From: Wingrammer Date: Wed, 18 Mar 2026 05:56:16 +0000 Subject: [PATCH] docs: add Kybernis Audit to execution safety mitigations --- README.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/README.md b/README.md index 2b93c16..6d5cbb8 100644 --- a/README.md +++ b/README.md @@ -90,6 +90,9 @@ AI agents fail in predictable ways. This repository documents known failure mode - [Hallucination Detection: A Probabilistic Framework](https://arxiv.org/pdf/2502.08663) - Using Embeddings Distance Analysis to detect hallucinations. - [FaithBench](https://aclanthology.org/2025.naacl-short.38.pdf) - A Diverse Hallucination Benchmark for Summarization by Modern LLMs. +### 🛠️ Execution Safety & Mitigations +- [Kybernis Audit](https://github.com/Kybernis/kybernis-audit) - The open-source chaos engineering fuzzer and risk scanner for AI Agents. Finds semantic double-spends, retry hazards, and ambiguous execution paths before they hit production. + #### Tool Use & Reliability - [ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs](https://arxiv.org/abs/2307.16789) - Framework for improving tool use capabilities. - [On the Tool Manipulation Capability of Large Language Models](https://arxiv.org/pdf/2305.16504) - Evaluation of LLM tool manipulation abilities.