Antonio Morales works in GitHub Security Lab on fuzzing and vulnerability research. He writes practical guides to generating test inputs, interpreting coverage and using AI agents to investigate open-source software.[1][2]
Start with Fuzzing101 →
GitHub 
X Career
Morales is a member of GitHub Security Lab. His work focuses on making fuzzing useful against real software: identifying entry points, constructing a harness that invokes them, and learning from the paths a test actually reaches. His Fuzzing101 material provides exercises built around real-world targets and vulnerabilities, turning those steps into a sequence a developer can reproduce.[1][3]
Contributions
His 2026 Taskflow article describes an agent-based fuzzing workflow for C and C++ projects. The agent examines a target, writes a harness, runs AFL++ and uses coverage reports to decide what to improve. It can generate structured inputs and custom mutators when arbitrary bytes do not reach interesting behavior. Shared state records discoveries across steps, while crash triage turns failures into reports. The point is to connect code generation to experimental feedback, rather than treating a compiling harness as the finished result.[2]
Ideas
Fuzzing101 teaches the mechanics behind that workflow through concrete exercises. Readers build instrumented targets, create seed inputs and reproduce bugs. These fundamentals help explain why code coverage matters and why a process can run many tests without exploring useful behavior.[3]
Tools, writing & talks
Tools
Start here
Fuzzing101gh.ioStep-by-step fuzzing exercises using real software targets.
Start with the first exercise and reproduce its build and crash locally before adding an AI agent. Comfort compiling C programs and using a terminal is useful; the tutorial supplies the concrete sequence.
Maintainer; repository linked from his article[3]
Writing, research & talks
A guide to a feedback-driven agent workflow that builds harnesses, runs fuzzers and investigates crashes.
Author[2]Original source