Building AI tools, desktop apps, and personal automation that can be used by real people on real machines.
Codex Quota Menubar · Desktop Pet · Parcel Pilot MCP · Notes
- AI-native personal tools: MCP servers, Codex helpers, and browser-session workflows that turn private, local data into useful AI actions.
- Desktop applications: small, shippable apps with downloads, release automation, GitHub Pages distribution, and macOS menu bar ergonomics.
- Engineering notes: repeatable debugging notes, GitHub Trending observations, and practical backend/frontend work logs.
| Project | What it is | Status |
|---|---|---|
| codex-quota-menubar | macOS menu bar utility for Codex 5-hour and 1-week quota visibility, with live-first account quota and local-log fallback. | Packaged macOS release with bilingual README and GitHub Actions build. |
| codex-pet-desktop | Windows/macOS desktop pet app with independent .petpack resources. |
Downloadable releases and GitHub Pages distribution. |
| parcel-pilot-mcp | Local-first MCP server for personal package tracking with Taobao, JD, and Cainiao browser sessions. | Early public release for personal-use workflows. |
| notes | Personal technical notes and daily GitHub Trending TOP3 observations. | Docsify site published with GitHub Pages. |
I am polishing projects that are small enough to install, inspect, and actually use: clear READMEs, runnable releases, visible screenshots, and honest safety boundaries. The current thread is mostly desktop utilities, local AI workflows, and notes that preserve what was learned while building them. If a repo is forked, it is usually kept as a learning/reference trail; the original projects above are the ones I am actively packaging for public use.