Kevin Stubbings works in GitHub Security Lab and writes about practical vulnerability research. His guides connect CodeQL analysis, browser security and AI-assisted audits with concrete examples developers can reproduce.[1][2]
Start with How we found 24 Android vulnerabilities using our open source AI security agent →
GitHub Career
Stubbings’s GitHub writing covers browser extensions, CORS configuration, DNS rebinding and SELinux. These subjects concern boundaries between components that should not automatically trust each other. His articles explain the vulnerable code patterns and, in the CORS and extension work, how CodeQL can help find them.[2]
Contributions
Stubbings’s Android research report explains targeted taskflows for GitHub Security Lab’s open-source Taskflow Agent. These prompts direct a model toward particular attack surfaces. The team reported 24 Android vulnerabilities, including WebView and exposed JavaScript-bridge issues, and published reproducible flows.[1]
Ideas
Stubbings reports that models recognize risky APIs and generate proofs of concept, but can misjudge exploitability in context. Misunderstanding a function’s returned data can create a false positive. A debugger or focused researcher follow-up can check the behavior. The report couples a reproducible audit workflow with limits on automated assessment.[1]
Tools, writing & talks
Writing, research & talks
Start here
A practical report on targeted Android taskflows, findings and limits of AI-assisted severity assessment.
Start with the two vulnerability examples, then follow the taskflow instructions if you know Android application structure. Treat the reported findings as a guide to investigation, with the article’s severity and false-positive discussion in view.
Author[1]Original source- Kevin Stubbings: security articlesgithub.blog
Guides to CORS, browser extensions, CodeQL and Linux access control.
Author archive[2]