π Shenzhen | π€ AI-native builder | π Indie Hacker
I like building interfaces that feel clean, fast, and intentional. The details matter: spacing, hierarchy, motion, and the little things that make a product feel finished.
- π§ memory-os β Local-first context layer for AI work. Profile, projects, insights β served via CLI, API, and MCP
- π chroniq β CLI-first personal logging for humans and agents. Append-only JSONL, one command to capture.
npm i -g @hazellin/chroniq - π©Ί mcp-skill-doctor β Pre-flight diagnostics for MCP servers. Catches PATH, env, and cwd issues before the protocol layer loads
- π bookmark-organizer β Audit, flatten, and restructure browser bookmarks without deleting anything
- 𧩠vue3-filter β Composable filter component for Vue 3 applications
The gap between AI capabilities and daily workflow is mostly an infrastructure problem. I build the small, sharp tools that close it:
- Context persistence β so AI doesn't start from zero every session
- Agent-friendly CLIs β structured output, local-first, zero lock-in
- MCP ecosystem tooling β making the last mile of server setup less painful
I also care about craft on the frontend: clean interfaces built with Vue, React, and TypeScript, where spacing, hierarchy, and motion all earn their place.
- Shipping memory-os as a reusable context layer across Claude, Codex, and terminal agents
- Growing mcp-skill-doctor from MVP to daily-driver reliability
- Making the final 10% of a UI feel easy instead of expensive
