I build tools that make AI coding workflows easier to inspect and use: cost tracking, instruction debugging, and plugin interfaces.
简体中文 · Projects · Open-source contributions
Understand where your AI coding spend goes. A DeepSeek Harness plugin for per-session costs, token usage, budgets, provider balances, and coding-plan quotas, with English and Chinese interfaces.
- Separates subscription usage from pay-as-you-go spending.
- Tracks model-level costs and historical usage, with configurable pricing.
- Developed around community reports and documented fixes.
Installation · Screenshots and usage · Changelog
See which AGENTS.md files are discovered, which overrides hide other files, and where a byte budget cuts off instructions. Includes a CLI and an interactive offline report, with no runtime dependencies.
npx codex-context-map . --html context-map.htmlTry the demo · Discovery rules and limitations · CI
| Project | What it does |
|---|---|
| DSH UI Hub | Finds plugin UI elements and lets you show, hide, move, resize, and arrange them. |
| DSH Sticky Disclosure | Collapses expanded thinking and tool sections with one action or a configurable shortcut. |
My current projects use JavaScript and Node.js, with browser interfaces and command-line tools. I work on reproducible bug reports, regression tests for fixes, clear setup instructions, and explicit behavior limits.
If you find a problem or have a concrete use case, open an issue in the relevant repository. A small example, expected behavior, and actual result make a useful starting point. English and Chinese are welcome.


