T-skills is a personal collection of eight skills for Claude Code and Codex. Each one drills a single engineering habit, sets a clear goal and the constraints that matter, then lets the model do what it does best. Every skill output starts with a 🐯 marker so you can tell at a glance which skill is running.
| Skill | When | What it does |
|---|---|---|
/think |
Before building anything new | Pressure-tests the design and produces a decision-complete plan. |
/design |
Building frontend interfaces | Produces distinctive UI with a committed direction, not generic defaults. |
/check |
After a task, before merging or release | Reviews the diff, extracts project constraints, handles release follow-through. |
/hunt |
Any bug, regression, or unexpected behavior | Systematic debugging; root cause confirmed before any fix. |
/write |
Writing or editing prose | Rewrites prose to sound natural in Chinese and English. |
/learn |
Diving into an unfamiliar domain | Six-phase research workflow: collect, digest, outline, fill in, refine, publish. |
/read |
Any URL or PDF | Fetches content with platform-specific routing; summary or clean Markdown. |
/health |
Auditing agent health | Checks Claude/Codex/Pi config, verifier output, and AI maintainability. |
Claude Code
npx skills add tanglei168/T-skills -a claude-code -g -yInstall just one with npx skills add tanglei168/T-skills --skill think -a claude-code -g -y.
Codex
npx skills add tanglei168/T-skills -a codex -g -yUpdate
npx skills update -g -yT-skills is derived from Waza by Tw93, used under the MIT License. The skill set, validation, codegen, and packaging machinery originate there; see LICENSE for attribution. Thanks to the original author.
MIT License. See LICENSE.