Skip to content

About

My GitHub profile - AI Agent Runtime & Execution Safety | 专注 AI Agent 运行时与执行安全

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Latest commit

 

History

29 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 

Repository files navigation

Hi, I'm Jason Cat 👋

更完整的个人介绍、开源贡献、实战项目和持续更新的学习笔记,都在 jasonbuildai.github.io。

Home · Open Source · Projects · Learning

个人主页 · 开源贡献 · 实战项目 · 学习空间

Focus: AI Agent Runtime & Execution Safety

专注方向:AI Agent 运行时与执行安全 —— 让 Agent 跑得起、跑得稳、跑得安全

Recent work 最近在做: loopx credential redaction & public-boundary safety · OpenSandbox atomic writes · openclaw streaming re-parse fixes · open-code-review diff parsing


📊 By the Numbers 数据速览

Rolling 12 months · 近 12 个月(截至 2026-11-16)

Contributions ¹ Commits PRs merged Issues
4,345 4,160 162 194

¹ GitHub contribution graph, trailing 12 months (as of 2026-11-16); commits, merged PRs and issues counted via GitHub search over the same window.

🚀 Selected Contributions 代表贡献

Project ⭐ What I did
openclaw/openclaw 394.6K Fixed streaming re-projection stalls across HTML, JSX, code-block, and list boundaries; contained tool-call text leaks.
修复 HTML、JSX、代码块与列表边界的流式重投影卡顿,并收敛工具调用文本泄漏。
bytedance/deer-flow 85.4K Added native Windows process spawning and PATH resolution; hardened search payloads, large inputs, and symlink tests.
实现原生 Windows Agent 启动与 PATH 解析,并增强异常搜索载荷、超大输入与 symlink 测试。
alibaba/open-code-review 52.6K Isolated CLI tests from the real home directory and fixed bare diff --git header splitting in diff parsing.
隔离 CLI 测试与真实 home 目录,并修复 diff 解析中裸 diff --git 头的切分问题。
agentscope-ai/QwenPaw 37.8K Made session-state persistence atomic and aligned channel configuration types with the built-in registry.
实现会话状态的原子化持久化,并使通道配置类型与内置注册表保持一致。
opensandbox-group/OpenSandbox 17.3K Hardened the Go, JS, Kotlin, and C# SDKs across tenant-key isolation, SSE wake retries, idempotent writes, file-transfer timeouts, and non-UTF-8 CLI output.
加固 Go、JS、Kotlin 与 C# SDK,覆盖租户密钥隔离、SSE 唤醒重试、幂等写入、文件传输超时与非 UTF-8 CLI 输出。
loopx-project/loopx 6.7K Added Windows machine identity for task leases, interrupted Goal recovery, UTF-8 report artifacts, CI and canary gates, and signed NSIS updates.
实现任务租约的 Windows 机器标识、中断 Goal 恢复、UTF-8 报告产物、CI 与 canary 门禁,以及 NSIS 签名更新。
apache/maka 6.0K Closed secret-redaction gaps for fine-grained GitHub PATs and shared redactors; added bounded Electron install retries in CI.
补齐细粒度 GitHub PAT 与共享脱敏器中的密钥脱敏漏洞,并为 CI 中的 Electron 安装增加限时重试。
the-open-agent/openagent 5.7K Added fallback-aware auto chat titles, resilient usage metadata, a Windows GUI (UIA) automation tool, and configuration fixes.
实现带回退的聊天标题自动生成、用量元数据容错、Windows GUI(UIA)自动化工具,并修复配置问题。
FireRedTeam/FireRed-OpenStoryline 3.7K Strengthened session-restore invariants, config-secret redaction, drag-and-drop uploads, typed errors, and retries.
强化会话恢复不变量、配置密钥脱敏、拖拽上传、类型化错误与重试机制。
0xsline/OpenChatCut 3.1K Isolated Windows verification suites; added Ctrl-chord shortcuts, whisper.cpp token grouping, and local media import.
隔离 Windows 验证套件,并实现 Ctrl 组合快捷键、whisper.cpp 分词与本地媒体导入。

… and 15+ more repositories — HiRoute, agentbench, casdoor, langchain-ai/deepagents and others. 另外还有 15+ 个仓库的贡献。

🔭 What I Focus On 专注方向

Direction What I work on
🧠 Agent Runtime & Reliability
运行时与可靠性
Session and goal lifecycle, scheduling, locks and leases, quotas, and streaming stability.
会话与目标生命周期、调度、锁与租约、配额和流式稳定性。
🛡️ Sandboxing & Execution Security
沙箱与执行安全
Sandbox lifecycle, egress and ingress policy, secret hygiene, and multi-language SDKs and CLIs.
沙箱生命周期、网络出入口策略、密钥卫生,以及多语言 SDK 与 CLI。
🖥️ Cross-platform Engineering
跨平台工程
Reliable agent tooling on Windows and WSL, with strong CI and verification gates.
让 Agent 工具在 Windows 与 WSL 上稳定可用,并建立可靠的 CI 与验证门禁。

🤝 Community & More 社区与日常

  • Member of @the-open-agent and @effectorHQ
    两个开源组织的成员
  • Long-term goal: keep making AI agents run fast, steady and safe in open source
    长期目标:持续让 AI Agent 在开源世界里跑得稳、跑得安全
  • Off the keyboard: football ⚽, exploring new places 🌍, meeting people with stories ✨
    代码之外:足球、到处走走、认识有故事的人
  • Always learning — there will always be people better than me, and that's a good thing.
    永远在学习——总有人比我更强,而这正是好事。

📫 Get in Touch 联系我

Email · GitHub · WeChat HJN522312MOST

Happy to talk about agents, sandboxes and runtime reliability — or just to say hi.
欢迎交流 AI Agent、沙箱与运行时可靠性——也欢迎单纯打个招呼。如果我的工作对你有帮助,一颗 ⭐ 或一个关注就是最大的鼓励。

🌐 Learn more at jasonbuildai.github.io
想了解更多?欢迎访问我的个人网站。

About

My GitHub profile - AI Agent Runtime & Execution Safety | 专注 AI Agent 运行时与执行安全

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors