<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI on Steven Lin's Blog</title><link>https://stevenlin.cc/blog/tags/ai/</link><description>Recent content in AI on Steven Lin's Blog</description><generator>Hugo -- gohugo.io</generator><language>en-US</language><lastBuildDate>Sun, 11 Oct 2026 00:00:00 +0800</lastBuildDate><atom:link href="https://stevenlin.cc/blog/tags/ai/index.xml" rel="self" type="application/rss+xml"/><item><title>Agent Steven</title><link>https://stevenlin.cc/blog/p/agent-steven/</link><pubDate>Sun, 11 Oct 2026 00:00:00 +0800</pubDate><guid>https://stevenlin.cc/blog/p/agent-steven/</guid><description>&lt;p&gt;&lt;strong&gt;Beyond Vulnerability Detection. Toward Attack Vector Discovery.&lt;/strong&gt;&lt;/p&gt;
&lt;h2 id="the-problem"&gt;&lt;a href="#the-problem" class="header-anchor"&gt;&lt;/a&gt;The Problem
&lt;/h2&gt;&lt;p&gt;AI is becoming increasingly effective at finding vulnerabilities in source code. It can identify insecure patterns, detect implementation flaws, and even help rewrite software in memory-safe languages like Rust.&lt;/p&gt;
&lt;p&gt;But finding more vulnerabilities is not the same as understanding how a product can be attacked.&lt;/p&gt;
&lt;p&gt;Some of the most significant security risks emerge not from individual bugs, but from &lt;strong&gt;architectural decisions, trust boundaries, implicit assumptions, and interactions between otherwise secure components&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;Traditional security scanners and AI-powered code analysis tools largely focus on isolated vulnerabilities. They often lack the broader system context needed to understand how attackers could exploit legitimate functionality or chain multiple components into unexpected attack paths.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;The next frontier of product security isn&amp;rsquo;t just finding more bugs. It&amp;rsquo;s discovering attack vectors at the system level.&lt;/strong&gt;&lt;/p&gt;
&lt;h2 id="what-were-building"&gt;&lt;a href="#what-were-building" class="header-anchor"&gt;&lt;/a&gt;What We&amp;rsquo;re Building
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Agent Steven&lt;/strong&gt; is an AI-powered product security initiative built around Claude, focused on discovering architecture-level attack vectors.&lt;/p&gt;
&lt;p&gt;Our goal is to connect fragmented engineering knowledge—including architecture diagrams, design documents, source code, API specifications, and historical security findings—into a unified understanding of a product&amp;rsquo;s security posture.&lt;/p&gt;
&lt;p&gt;Rather than simply asking &lt;em&gt;&amp;ldquo;Where is the vulnerability?&amp;rdquo;&lt;/em&gt;, we want AI to reason about more fundamental questions:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;What are the actual entry points and trust boundaries?&lt;/li&gt;
&lt;li&gt;What assumptions do components make about one another?&lt;/li&gt;
&lt;li&gt;Can individually secure components create an insecure system when combined?&lt;/li&gt;
&lt;li&gt;What attack paths might exist that traditional vulnerability scanners cannot identify?&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="our-approach"&gt;&lt;a href="#our-approach" class="header-anchor"&gt;&lt;/a&gt;Our Approach
&lt;/h2&gt;&lt;p&gt;We&amp;rsquo;re exploring a multi-agent security analysis pipeline powered by Claude:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Knowledge Integration&lt;/strong&gt; — Connect distributed engineering knowledge and product documentation.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Architecture Understanding&lt;/strong&gt; — Build a system-level model of components, interfaces, trust boundaries, and dependencies.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Attack Vector Discovery&lt;/strong&gt; — Reason across components to identify overlooked attack surfaces, unsafe architectural assumptions, and potential multi-step attack paths.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Security Validation&lt;/strong&gt; — Generate evidence-backed security hypotheses and actionable threat models for human review.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The goal is to complement existing vulnerability detection—not replace it.&lt;/p&gt;
&lt;h2 id="why-claude"&gt;&lt;a href="#why-claude" class="header-anchor"&gt;&lt;/a&gt;Why Claude?
&lt;/h2&gt;&lt;p&gt;Understanding complex systems requires more than analyzing individual code snippets.&lt;/p&gt;
&lt;p&gt;We see Claude&amp;rsquo;s reasoning, long-context capabilities, and agentic workflows as promising building blocks for connecting fragmented engineering knowledge and investigating security implications across entire product architectures.&lt;/p&gt;
&lt;p&gt;Our longer-term vision is to make architecture-level security reasoning a continuous part of the &lt;strong&gt;Secure Development Lifecycle (SDL)&lt;/strong&gt;, rather than an occasional manual review.&lt;/p&gt;
&lt;h2 id="project-status"&gt;&lt;a href="#project-status" class="header-anchor"&gt;&lt;/a&gt;Project Status
&lt;/h2&gt;&lt;p&gt;Agent Steven is an early-stage, independently developed project exploring AI-driven product security research and automation.&lt;/p&gt;
&lt;p&gt;Our initial focus is on building and evaluating a practical workflow for knowledge integration, architectural reasoning, and attack-vector discovery.&lt;/p&gt;
&lt;p&gt;All security findings are intended to be validated through human review and authorized security testing.&lt;/p&gt;
&lt;h2 id="about-the-founder"&gt;&lt;a href="#about-the-founder" class="header-anchor"&gt;&lt;/a&gt;About the Founder
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Steven Lin&lt;/strong&gt; is a Product Security Engineer with experience across application, embedded system, and hardware security.&lt;/p&gt;
&lt;p&gt;His background includes vulnerability research, security testing, reverse engineering, and integrating security practices into the software development lifecycle.&lt;/p&gt;
&lt;p&gt;Agent Steven brings these experiences together to explore how AI can move product security beyond individual vulnerability detection toward deeper system-level understanding.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Contact:&lt;/strong&gt; &lt;a class="link" href="mailto:me@stevenlin.cc" &gt;me@stevenlin.cc&lt;/a&gt;&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;LinkedIn:&lt;/strong&gt; &lt;a class="link" href="https://www.linkedin.com/in/5teven1in" target="_blank" rel="noopener"
&gt;Steven Lin&lt;/a&gt;&lt;/p&gt;</description></item></channel></rss>