---
title: "austine omondi (@austoo) | Hashnode"
description: "AI Security | AI Engineer. Read the latest articles by austine omondi on Hashnode."
url: "https://hashnode.com/@austoo"
---

@austoo

# austine omondi

@austoo · Nairobi, Kenya · Joined June 2026

AI Security | AI Engineer

### [A Practical Methodology for Pentesting RAG Applications](https://blogaustinomondicom.hashnode.dev/a-practical-methodology-for-pentesting-rag-applications?utm_source=hashnode&utm_medium=feed)

A field-tested checklist and playbook drawn from 12 weeks of lab work across three deliberately vulnerable applications. What this is This post is a practical methodology for pentesting Retrieval-Aug

23 min read · [blogaustinomondicom.hashnode.dev](https://blogaustinomondicom.hashnode.dev/a-practical-methodology-for-pentesting-rag-applications?utm_source=hashnode&utm_medium=feed)

[Aug 8](https://hashnode.com/posts/a-practical-methodology-for-pentesting-rag-applications/6a774a4c9772bab572d1dfa8)

### [Eight Ways to Break a Multi-Agent System](https://blogaustinomondicom.hashnode.dev/eight-ways-to-break-a-multi-agent-system?utm_source=hashnode&utm_medium=feed)

What I learned building a deliberately vulnerable LangGraph system and attacking it Most prompt-injection writing treats the LLM as a single box: one prompt in, one completion out, one place for the a

11 min read · [blogaustinomondicom.hashnode.dev](https://blogaustinomondicom.hashnode.dev/eight-ways-to-break-a-multi-agent-system?utm_source=hashnode&utm_medium=feed)

[Jul 13](https://hashnode.com/posts/eight-ways-to-break-a-multi-agent-system/6a5509f52e3f7b4754e525cf)

### [MyLLMBank — Prompt Injection Against a ReAct Banking Agent](https://blogaustinomondicom.hashnode.dev/myllmbank-prompt-injection-against-a-react-banking-agent?utm_source=hashnode&utm_medium=feed)

Target: MyLLMBank (Reversec HackAI challenge) — a Streamlit ReAct agent over OpenAI GPT-4 Turbo, exposing two tools (GetCurrentUser, GetUserTransactions) against a transactions database. Objectives: t

10 min read · [blogaustinomondicom.hashnode.dev](https://blogaustinomondicom.hashnode.dev/myllmbank-prompt-injection-against-a-react-banking-agent?utm_source=hashnode&utm_medium=feed)

[Jul 1](https://hashnode.com/posts/myllmbank-prompt-injection-against-a-react-banking-agent/6a451fa32070d6e29be6ba92)

### [The Secret Was Never the Hard Part](https://blogaustinomondicom.hashnode.dev/the-secret-was-never-the-hard-part?utm_source=hashnode&utm_medium=feed)

Breaking three different output-side guardrails in the HackAI and MyLLMBank labs — and what each failure teaches a defender. Most LLM jailbreak write-ups end at the winning prompt. This one is about

12 min read · [blogaustinomondicom.hashnode.dev](https://blogaustinomondicom.hashnode.dev/the-secret-was-never-the-hard-part?utm_source=hashnode&utm_medium=feed)

[Jul 1](https://hashnode.com/posts/the-secret-was-never-the-hard-part/6a44c518fe531d1d16f38666)

### [State Pollution in Multi-Agent LangGraph Systems: When One Agent Can Silently Reprogram the Others](https://blogaustinomondicom.hashnode.dev/state-pollution-in-multi-agent-langgraph-systems-when-one-agent-can-silently-reprogram-the-others?utm_source=hashnode&utm_medium=feed)

Most conversations about LLM security still stop at the chatbot. Prompt injection, jailbreaks, secret extraction — well-understood attack surfaces, mostly on single-turn systems where one model reads

8 min read · [blogaustinomondicom.hashnode.dev](https://blogaustinomondicom.hashnode.dev/state-pollution-in-multi-agent-langgraph-systems-when-one-agent-can-silently-reprogram-the-others?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/state-pollution-in-multi-agent-langgraph-systems-when-one-agent-can-silently-reprogram-the-others/6a44c4353344dedab7380b7d) [Jul 1](https://hashnode.com/posts/state-pollution-in-multi-agent-langgraph-systems-when-one-agent-can-silently-reprogram-the-others/6a44c4353344dedab7380b7d)