RSS Amplifier

Blog

Agentic AI

I share my thoughts on anything related to Agentic AI and Agentic AI Security topics. You can also subscribe to our sister YouTube channel at https://www.youtube.com/@AgenticAISignal/shorts for the latest signal on Agentic AI

kenhuangus.substack.comSource feed ↗11 posts

Live Last read · last published · next check

Written by

Latest posts

From Software Engineering to Harness Engineering: What OpenAI's Codex Actually Builds

On August 19, 2026, OpenAI published Codex as a platform: build on the open agent harness, emphasizing that the most valuable reusable asset inside Codex is not the chat interface.

FOR IMMEDIATE RELEASE: OWASP Publishes Agentic Skills Top 10 v1.0

A community-built security standard for the agent skill ecosystem, mapping ten risk classes to AISVS, ASI, MCP, and LLM controls

Harness Engineering: Design Patterns for Securing Long-Horizon Multi-Agent AI Systems

The professional edition of Harness Engineering is now available on Kindle.

40% OFF: Hands-On Harness Engineering with Claude

Registration is live for Hands-On Harness Engineering with Claude—and you can get 40% off using our exclusive discount code.

Black Hat and DEF CON 34 Trip Report: What Vegas Actually Shipped in Agentic AI Security

Twenty vendors said agentic. I checked every claim against primary sources. What is real, what was only a slide, and where the category still has a hole.

Coding Agent Security: Lessons from Claude Code, Cowork, Codex, and Copilot in the Wild

This article is sponsored by Reva AI (reva.ai).

Inside the 1-Bit LLM: How Bonsai Fits a 27B Model on a Phone

In July 2026, PrismML shipped a 27-billion-parameter language model, built on Alibaba's Qwen3.6-27B, that runs in a browser tab on WebGPU and, separately, on a phone.

Part 1: A Software Factory Agentic Skill for Greenfield Builds

This is Part 1 of a two-part software-factory series.

MAESTRO Analysis of OpenAI and Anthropic Agent Hacking Incidents

Two evaluation escapes in one week. Mapped onto the seven MAESTRO layers, one is an operations failure and the other is an alignment failure, and the fix lists barely overlap.

One Week, Three Conferences: My Las Vegas AI Security Itinerary

I’ll be in Las Vegas Aug 3–10 for Ai4, Black Hat, and DEFCON.

Harness Engineering as the Umbrella Discipline

from software engineering to harness engineering...