name : L1ngSh1 (Yuze_Yang)
role : Security Engineer / Researcher
focus : Agent Security · Program Analysis · Vulnerability Research · Reverse
major : Cyber Security · Computer Science · Software Engineering · Electrical Engineering
building : Security evaluation and automated analysis systems
research : Indirect prompt injection and trustworthy AI agents（2026.6）
status : LEARNING · BUILDING · BREAKING
I build security tools, reproduce research systems, and study how intelligent agents fail under adversarial conditions.
01 — Agent Security
Indirect prompt injection, tool-use risks, agent isolation, and adversarial evaluation.
02 — Program Analysis
Static analysis and source-code understanding with compiler and AST infrastructure.
03 — Vulnerability Evaluation
Reproducible benchmarks and metrics for security tools and automated vulnerability discovery.
04 — Offensive Security
Practical red-team skills grounded in real attack surfaces and defensible research questions.
A vulnerability-evaluation platform for analysis systems. Measures and compares automated security tools through reproducible experiments, explicit metrics, and reviewable evidence.
Security Evaluation · Benchmark · Vulnerability Detection
Program-analysis infrastructure for source-level reasoning. Built with C++ and Clang AST, with a focus on templates, namespaces, class hierarchies, initialization, and explicit source relationships.
C++ · Clang · LLVM · AST · Program Analysis
A reproducible environment for indirect prompt-injection research. Evaluates attacks against LLM agents and tool-using systems inside controlled Docker-based experiments.
Agent Security · Prompt Injection · Docker · Android
ctf-lab— practical CTF exercises and offensive-security methodology.netwatch-cli— lightweight network and system inspection for security-engineering workflows.
[ACTIVE] Agent security and indirect prompt injection
[BUILDING] Program-analysis infrastructure
[TESTING] Reproducible vulnerability evaluation
[READING] Security top-conference research
[TRAINING] Offensive security and CTF methodology
01 Build systems, not demos.
02 Measure security claims with reproducible evidence.
03 Understand how systems break before claiming they are safe.
