I'm an Agent builder working on systems that stay close to users, their tools, and their data.
我在构建真正能长期生活在用户身边的 AI Agent：本地优先、可审计、可组合，并且尊重人的控制权。
My current questions are less about “how smart is the model?” and more about:
- How should an Agent connect to personal data without creating another cloud silo?
- How can memory become structured, queryable, and traceable instead of opaque context?
- How do MCP tools, skills, and runtimes compose into a reliable everyday system?
- How can powerful automation still keep consent and human judgment in the loop?
| System | What I'm exploring |
|---|---|
| 🫀 Vital Agent Sync | A local-first Apple Health connector that exposes user-controlled, scoped context to MCP-compatible Agents. |
| 🧠 MemTable | A structured ledger for Agents: confirmed records, timestamps, schemas, aggregation, and source tracing — not another vector-memory wrapper. |
| 📚 Oh My Notion | A local-first Notion Agent with hybrid retrieval, skill routing, grounded answers, and inspectable memory. |
Memory recalls what happened. A ledger computes what changed.
local over opaque
inspectable over magical
composable over monolithic
consent-aware over always-on
useful over impressive
- Building Agent-first onboarding and MCP infrastructure in Vital Agent Sync
- Contributing fixes and product improvements to Hermes Agent
- Experimenting with skills, memory, local runtimes, and human-in-the-loop workflows
Blog · X / Twitter · GitHub
Building Agents that grow useful without growing beyond their users' control.