English · 简体中文
Frontend / Full-stack developer working on enterprise applications, open-source virtualization, design systems, and multi-agent development.
- 🤖 Building workflows where Codex, Claude Code, and other coding agents can collaborate with durable context and clear ownership.
- 🧭 Exploring multi-agent orchestration with goals, delegation, governance, budgets, approvals, and auditable execution.
- 🧪 Treating prompts, context, MCP tools, and evals as versioned engineering assets rather than disposable chat history.
- 🔍 Designing agent-assisted development around observability, traceability, reproducibility, and root-cause debugging.
- 🏗️ Applying AI-native workflows to complex enterprise software, virtualization, disaster recovery, and open-source products.
| Area | Work |
|---|---|
| AI Agent Systems | Multi-agent coordination, persistent task context, runtime skills, governance, cost control, and execution audit trails. |
| AI-Native Engineering | Repository context systems, prompts, MCP integrations, eval suites, agent definitions, and reproducible development workflows. |
| Enterprise Frontend | Large React and TypeScript applications, design systems, micro-frontends, Module Federation, and Nx monorepos. |
| Cloud & Virtualization | Management consoles, documentation platforms, recovery workflows, observability, and deployment automation. |
| Open Source Platforms | Product websites, documentation systems, community infrastructure, bilingual content, search, and release validation. |
Nx · Rspack · Module Federation · Playwright · OpenAPI · Monorepo · MCP · AI Evals
- Fail fast — errors should never pass silently.
- Fix causes, not symptoms — avoid papering over bugs.
- Make it observable — critical paths need useful logs, metrics, and traces.
- Design for debugging — important decisions and state transitions should be explainable.
- Keep documentation alive — code, contracts, and operational knowledge evolve together.
- Protect the mainline — isolate risky work and keep changes reviewable.