WEB SECURITY RESEARCH
Web Security
Attack surfaces, anti-debugging, code protection and the less obedient corners of the browser.
ACCESS / JC-07 SYSTEM ONLINE
IDENTITY VERIFIED · CLEARANCE GRANTED
JAY CHIU
FRONTEND ENGINEER / SECURITY EXPLORER
Building interfaces with code, curiosity and a little controlled chaos.
NODE 07SHANGHAI / CNLATENCY 12MS
IDENTITY TOKEN LOADING
↔ DRAG / CLICK TO FLIP
02 / CURRENT WORK
Current work
The active projects, ongoing research topics and collaboration status shaping my work right now.
FEATURED PROJECTVIEW ON GITHUB ↗RESUMECVFrontend EngineerSECURITY LAB CREATIVE SYSTEMS
Open-source Code Agent workspace
OhMyCode
One runtime for local tools, persistent terminals, recoverable execution and Multi-Agent collaboration.
View professional experience, selected projects and technical skills.
OPEN PDF ↗VIEW RELATED RECORD ↗
ONGOING PRACTICE
Frontend / AI / Interaction
Combining engineering, visual systems and experimental interaction into memorable digital experiences.
03 / ACTIVITY LOGS
VIEW ALL 12 RECORDSWriting archive
01OPEN ↗02OPEN ↗03OPEN ↗04OPEN ↗05OPEN ↗
我为什么又写了一个 Code Agent？OhMyCode 的设计与实现
从本地终端、文件工具和可恢复执行，到上下文压缩与 Multi-Agent，记录 OhMyCode 从聊天界面逐渐长成 Agent 工作空间的过程。
让大模型自己写量化策略，最后能赚钱吗？
从结构化策略 DSL、AI Agent 自动迭代到跨标的与跨时间回测，记录 Stock Broker AI 量化策略研究平台的设计与实现。
浏览器的阅读模式：Readability 正文提取算法解析，简单实现与改进
Readability 是一个可以提取网页正文的算法，它可以将网页中的广告、导航、评论等非正文内容过滤掉，只保留正文内容。
Tensorflow.js（四）：基于 node 使用长短期记忆网络进行文本分类
本文将介绍如何使用长短期记忆网络（LSTM）进行文本分类任务，LSTM 是一种特殊的 RNN，能够学习长期依赖关系，适合处理时间序列数据。
股价与股民情绪是否有相关性？基于大模型来快速验证
几年前就提出的假设，无奈难以实施，而现在大模型的出现让批量快速评估变得可行。
“I combine technology, visual systems and interaction into things that are useful and memorable.”
GITHUB / @Covfefeable ↗- ROLE
- Frontend Engineer
- LOCATION
- Shanghai, China
- FOCUS
- Security / AI / Creative Code
- EXPERIENCE
- Meizu / FreeBuf
- GITHUB
- @Covfefeable ↗
YOUR VISIT HAS BEEN LOGGED
Want to build something interesting?
START A CONVERSATION ↗SESSION ID: GUEST-2026STATUS: ACTIVE