Understanding RAG: Build a Security Assistant with Cloud Run, Firestore, and Vertex AI
Build a RAG-powered security assistant on Google Cloud using Firestore and Vertex AI — deployable threat intelligence with context-aware responses.
AI
A collection of 2 Posts
From Zero to Amazon Bedrock: Learn how to interact with GenAI Models using AWS Lambda
A hands-on Python guide to Amazon Bedrock with AWS Lambda — call foundation models, use Converse API, and apply minimal IAM to ship GenAI features fast.
AI-Agents
A collection of 2 Posts
One Prompt. Claude Fixed My Broken SIEM Feed.
A real-world POC: I gave Claude access to Elastic SIEM, Jira, Confluence, and Gmail — and let it triage a broken log feed autonomously. Here's exactly what happened, and how to set it up yourself.
Build Your First AI Agent: From Static AI to Autonomous Systems
Build your first autonomous AI agent with CrewAI — deployable Python code, multi-agent systems, and real API integrations for security engineers.
AI-Defense
A collection of 2 Posts
Build Your AI Defense #2: Check What Your Agent Does, Before It Does It
A free lab you run on your own device: an AI agent, a gateway in front of it, eight checks and four versions of one attack. Which checks stopped the agent before it mailed AWS keys to an attacker, which ones only looked like they did, and what each one costs. With a two-minute summary of the risk and how to control it.
Build Your AI Defense #1: Prompt Scanning Is Not Enough
An AI agent asked to summarize your inbox tries to send your password to an attacker. To stop that, AI security must check what agents do, not only what they say. Inside: the six areas of AI security, runtime security in depth, a checklist and ten vendor questions.
AI-Security
A collection of 4 Posts
I Built a Claude Skill to Summarize YouTube. So Did the Attackers.
My first Claude skill could install software on your computer without asking you. Then my antivirus said Claude had written me a Python stealer. What happened, how a skill runs things on your machine, and what to check before you install one. With a two-minute summary for leaders.
Learn by Doing · AI Security #3: Agent Hijacking
Lab #2 tricked an AI into saying a secret. Lab #3 tricks an agent into doing something: reading your inbox, finding a password, and sending it to an attacker, all from one poisoned email. The EchoLeak pattern, on your laptop.
Learn by Doing · AI Security #2: MCP Tool Poisoning
Lab #1 tricked a chatbot into leaking a secret. Now we give that chatbot tools through an MCP server and watch it get hijacked two new ways — a poisoned tool description, and instructions hidden in a tool's response. Build it, break it, understand it. All local, no API key.
Learn by Doing · AI Security #1: Prompt Injection
AI is outpacing the security around it. This series is a place for security people to learn by doing. Lab #1: build a vulnerable chatbot on your laptop, trick it into leaking a secret with one sentence, and understand why the usual guards fail.
Cloud-Security
A collection of 1 Post
Reduce MTTD/MTTR with Amazon Bedrock - From Telemetry to Action
Reduce MTTD/MTTR by combining AWS WAF, CloudWatch, and Amazon Bedrock Agents — from telemetry anomaly detection to automated security mitigation.
Detection-Engineering
A collection of 2 Posts
Hardening Azure Feeds Into Google SecOps
A leaked Azure storage key or Event Hub connection string is, by default, usable from anywhere on the public internet. Here's how to put that credential behind a network door — using the Azure firewall and Google's published goog.json IP ranges — with two parallel walkthroughs for the two architectures you'll actually find in the wild, plus a self-maintaining reconciler.
The Feed Health Monitor I'd Actually Deploy
A small Python service for Google SecOps that watches every feed, runs three independent checks per feed, restarts what it can, files one Jira ticket per outage with a Gemini summary, and emails the owner. No agent framework. No MCP. Just a YAML config and a Cloud Run Job. Open source, v1, contributions welcome.
Findings
A collection of 1 Post
I Built a Free Tool to Check Claude Skills. It Found a Gap in a Microsoft AI Plugin.
Two days after I reported it, Microsoft fixed it. What my free tool found in an official Microsoft AI plugin, why it matters even when nothing is malicious, how I reported it, and what to check in your own plugins. With a two-minute summary for leaders.
Prompt-Engineering
A collection of 1 Post
How to Talk to AI: Human–Computer Communication in the LLM Era
A practical field guide to prompting for security engineers — use structure, context, and examples to get consistent, high-quality results from LLMs.