Skip to content
View Jinx-1120's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report Jinx-1120

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Jinx-1120/README.md

Jinx

I build AI systems that can hold state, survive failure, and finish the job.

agent runtimes · workflow engineering · developer tools · applied research

I started with web interfaces and build tooling. These days I work closer to the boundaries between agents, runtimes, data, and people—mostly in TypeScript, with Rust and Python where the system calls for them.

I care about making AI work operational: explicit state, recoverable execution, human checkpoints, useful observability, and evidence that the real path works.

Recent public work

Project What I am exploring
Agent Skills Portable, evidence-first workflows for coding agents—designed around clear outcomes, adaptive depth, and truthful completion.
Ankole Contributing runtime, console, and cross-runtime E2E work to an open-source, self-hosted AgentOS for shared AI colleagues.
MDX Renderer A browser-based MDX playground with isolated preview, custom components, local persistence, and a deliberately small runtime.
agent-cli An experiment in multi-agent software workflows with durable state, DAG scheduling, human checkpoints, and breakpoint recovery.

How I work

inspect reality → model state & failure → ship a focused slice → verify the boundary → document the contract
  • Real runtime evidence beats plausible-looking output.
  • State, failure, recovery, and cancellation are product work—not cleanup.
  • Tests, docs, config, and examples should ship with the behavior they describe.
  • Repetition is usually a signal to build a tool, adapter, or reusable workflow.
  • I prefer small, descriptive commits and progressive hardening over one opaque rewrite.

Current toolbox

  • Languages: TypeScript · Rust · Python · SQL
  • Frontend: React · Next.js · Vite · Tailwind CSS · shadcn/ui · React-Native · Flutter
  • Backend & data: Bun · Node.js · Hono · Drizzle ORM · PostgreSQL · ClickHouse · Redis · BullMQ · Polars
  • AI & interfaces: AI SDK · MCP · oRPC · TanStack Query · Zod · hermes-agent
  • Engineering: Turborepo · Biome · Vitest · Bun Test · Rstest · Docker · Kubernetes · Helm · OpenTelemetry · GitHub Actions

If you are working on durable agents, developer infrastructure, or better human–agent workflows, I am interested in the hard parts: ownership, recovery, and proving that the system actually works.

Pinned Loading

  1. skills skills Public

    Open-source workflow skills for Codex-style coding agents.

    1 1

  2. mdx-renderer mdx-renderer Public

    Browser-based MDX playground with live preview, built-in components, and Tailwind CSS support

    TypeScript 1

  3. NervJS/taro NervJS/taro Public

    开放式跨端跨框架解决方案,支持使用 React/Vue 等框架来开发微信/京东/百度/支付宝/字节跳动/ QQ 小程序/H5/React Native 等应用。

    TypeScript 37.6k 4.9k

  4. yorkie/react-native-wechat yorkie/react-native-wechat Public archive

    🚀 WeChat login, share, favorite and payment for React-Native on iOS and Android platforms (QQ: 336021910)

    Objective-C 2.9k 706