Become a sponsor to Zhongsheng Ji

Hi, I’m Zhongsheng Ji (@Wh1isper), an AI engineer based in Shenzhen. My long-term focus is general agents: not just a chatbot loop or a narrow task demo, but systems that can work across models, tools, environments, memory, permissions, human feedback, and long-running tasks—and remain understandable and dependable while doing so.

Pydantic AI is an important foundation for this work, and I am an active upstream contributor. My merged contributions span history processors and RunContext, richer ToolReturn and deferred tool results, MCP tool namespacing, human-in-the-loop flows, provider and usage compatibility, Anthropic prompt caching, and new model, reasoning, and cache behavior. I also contribute to the surrounding ecosystem, including genai-prices, AnyCap, and production-oriented agent SDK and environment projects.

I turn those ideas into open-source systems of my own:

• YA / YA Claw explores a Pydantic AI–based general-agent stack: SDKs, environment abstractions, CLI and TUI layers, workspace-native runtime, schedules, bridges, and long-running agent services.
• AnyCap gives agents a capability runtime for media, live-web research, human feedback, collaboration, and delivering results people can actually use.
• Starweaver explores the same infrastructure questions in Rust through an agent SDK, CLI, RPC layer, desktop foundation, explicit boundaries, and reliable execution.
• mcp-email-server brings agents into a real-world workflow with secure, practical email access over MCP.

I also write about agent architecture and publish a weekly AI industry digest on my blog. Building in public helps me turn implementation experience into reusable code, clearer abstractions, and honest technical discussion.

Sponsorship gives me more time and infrastructure to maintain upstream contributions and independent projects, improve documentation and onboarding, strengthen cross-platform testing and security, respond to users, and pursue longer-term general-agent research that is difficult to sustain through short product cycles.

If you use my work, learn from my writing, or share my belief that general-agent infrastructure should be open, practical, and reliable, your support helps me keep building it. Thank you.