🇺🇸 English | 🇨🇳 简体中文

Build better AI coding behavior with production-oriented skills.

wayne-skills is a curated skill pack for AI coding agents (Codex/Claude style), centered on the pywayne ecosystem.
It converts practical modules into reusable SKILL.md playbooks so agents can pick better APIs, workflows, and constraints with less prompting.

• 🧭 Source-aligned: skills map directly to real module paths
• 🧱 Stable conventions: naming and structure stay predictable
• 🛡️ Lower-risk outputs: fewer ad-hoc decisions in code generation
• 🚀 Broad coverage: CV, VIO, DSP, LLM, statistics, automation, integrations

• pywayne/ - skills aligned with pywayne source modules
• alapi/ - full ALAPI OpenAPI skill with endpoint routing and request helpers
• send-email/ - SMTP email skill with templates and attachments
• deep-think/ - deep analysis and decomposition workflow
• shell-shortcuts/ - cross-platform terminal shortcut commands (proxy_on, goto, gpu)
• tutor-general/ - general tutoring skill for generating educational videos with Manim
• tutor-math-geometry/ - math geometry tutoring skill with interactive animations
• proactive-agent/ - proactive agent architecture (WAL, Working Buffer, etc.)
• week-report-system/ - AI-powered weekly work report system with Git-based conversation logging
• CLAUDE.md - naming, structure, and documentation rules
• AGENTS.md - concise instructions for external models/agents

source module path -> skill directory -> skill name

Examples:

• pywayne/llm/chat_bot.py -> pywayne/llm/chat-bot/ -> pywayne-llm-chat-bot
• pywayne/vio/SE3.py -> pywayne/vio/se3/ -> pywayne-vio-se3
• pywayne/cv/apriltag_detector.py -> pywayne/cv/apriltag-detector/ -> pywayne-cv-apriltag-detector

• alapi: full ALAPI OpenAPI skill with endpoint routing, auth guidance, and request helpers
• deep-think: structured deep reasoning workflow
• send-email: SMTP sending with HTML templates and attachments
• shell-shortcuts: set up proxy_on/proxy_off/goto/gpu and optional Conda autostart
• tutor-general: general tutoring skill with Manim video generation
• tutor-math-geometry: math geometry tutoring skill with interactive animations
• proactive-agent: proactive agent architecture (WAL, Working Buffer, etc.)
• week-report-system: AI weekly work report system — auto-records every conversation to GitHub and generates structured weekly reports on demand

Automatically turn your daily AI conversations into polished weekly reports.

week-report-system silently logs every user–AI exchange (compressed + summarized) to a private GitHub repository, organized by year and week. At any point, ask it to generate a full weekly report — it reads the conversation logs, categorizes work by project, extracts key metrics, and outputs a structured markdown report.

Repository structure:

week-reports/
├── 2026/
│   └── week12/
│       ├── 20260318-a1b2c3d4.txt      # brief conversation log
│       ├── 20260319-e5f6g7h8.txt
│       └── report-20260320-143022.md  # generated weekly report

How to use:

• Just chat normally — conversations are recorded in the background
• Say 写周报 / generate week report to get a full report
• Say 总结2026年第12周的工作 to report on a specific week

Install across all AI agents on your machine with Skill Manager:

npm -g install @wang121ye/skillmanager
skillmanager install --global   # select week-report-system from the list

One install, works for Claude, ChatGPT, and any other agent that supports skills.

• 🛠️ Developer Tools: tools, helper, bin/*, crypto
• 📊 Data & Math: dsp, maths, statistics, data-structure, plot
• 🤖 Vision & Robotics: cv/*, vio/*, calibration/*, ahrs/*, visualization/*
• 🔌 Platform & Integration: adb/*, cross-comm, aliyun-oss
• 💬 Product Interfaces: llm/*, lark-*, tts, gui

For full details, browse all **/SKILL.md files.

When adding/updating skills:

1. Follow naming rules in CLAUDE.md
2. Keep directory names in hyphen-case
3. Keep language links valid (README.md <-> README_ch.md)
4. Sync counts and domain text with actual SKILL.md files

MIT. See LICENSE.