Kembali ke skill
proagents
oleh Arlandaren
Windsurf—Belum ada penilaian
Penilaian Anda
Ringkasan
A curated skillstack for AI agents. Make your AI code editor write code like a Tech Lead. Stop copy-pasting instructions. Install 794 expert rules, workflows, and custom personas in Cursor, Claude Code, and Windsurf with one CLI command.
Deskripsi
<p align="center"> <img src="assets/banner.png" alt="proagents" width="100%"> </p> <p align="center"> <b>English</b> · <a href="README.ru.md">Русский</a> · <a href="README.zh.md">简体中文</a> </p> <p align="center"> <img src="https://img.shields.io/badge/total_prompts-794-6366f1?style=for-the-badge" alt="794 prompts"> <img src="https://img.shields.io/badge/personas-232-8b5cf6?style=for-the-badge" alt="232 personas"> <img src="https://img.shields.io/badge/workflows-521-a855f7?style=for-the-badge" alt="521 workflows"> <img src="https://img.shields.io/badge/license-MIT-22c55e?style=for-the-badge" alt="MIT"> </p> <p align="center"> <img src="https://img.shields.io/badge/Cursor-compatible-blue?style=flat-square" alt="Cursor"> <img src="https://img.shields.io/badge/Claude_Code-compatible-orange?style=flat-square" alt="Claude Code"> <img src="https://img.shields.io/badge/Windsurf-compatible-cyan?style=flat-square" alt="Windsurf"> <img src="https://img.shields.io/badge/Zed-compatible-green?style=flat-square" alt="Zed"> <img src="https://img.shields.io/badge/Gemini_CLI-compatible-yellow?style=flat-square" alt="Gemini CLI"> </p> --- **proagents** is a curated, prefix-free library of **794 professional system prompts, agent personas, and task workflows** distilled from 6 leading open-source agent frameworks. No wrappers, no telemetry, no dependencies. Drop any file into your AI IDE and the model immediately operates as a senior specialist. --- ## 📊 By the Numbers <table> <tr> <td align="center"><b>794</b><br><sub>Total prompts</sub></td> <td align="center"><b>232</b><br><sub>Agent personas</sub></td> <td align="center"><b>521</b><br><sub>Workflow checklists</sub></td> <td align="center"><b>41</b><br><sub>Rules & taste tokens</sub></td> <td align="center"><b>11</b><br><sub>Domains covered</sub></td> <td align="center"><b>6</b><br><sub>Source frameworks</sub></td> </tr> </table> --- ## ⚡ Why proagents? Most AI IDE prompts are written once, never maintained, and produce generic outputs. These were battle-tested across real production projects and distilled into a clean, domain-organized format. The difference is immediate: | | Default AI | proagents | |---|---|---| | **UI Output** | Generic Tailwind, boring layout, system fonts | Curated HSL palettes, micro-animations, glassmorphism | | **Code Safety** | Hardcoded values, missing edge cases | Zero magic strings, strict validation, typed errors | | **Testing** | "I'll add tests later" | TDD first: red → green → refactor | | **Security** | OWASP basics ignored | Threat modeling built into every review | | **Performance** | Unoptimized, re-renders everywhere | LCP targets, spring physics, precise DOM ops | --- ## 🚀 Quick Start ```bash git clone https://github.com/Arlandaren/proagents.git cd proagents # List all domains and counts ./proagents list # Search by keyword ./proagents search react # Install to Cursor (.cursorrules) ./proagents install react-patterns --cursor # Install to Claude Code (CLAUDE.md) ./proagents install senior-fullstack --stdout >> CLAUDE.md # Preview any prompt ./proagents install ux-architect --info ``` No pip installs. Works on Python 3.8+. --- ## 🗂️ Domain Overview ### 🧑💻 Engineering Personas — 126 agents Deep specialists, not generic "act as a developer" prompts. Each has a defined personality, strict rules, and measurable output standards. | Agent | What it does | |---|---| | `senior-fullstack` | Full-stack engineer with strict architectural opinions and code review standards | | `react-build-resolver` | Diagnoses and fixes Webpack/Vite build failures in React/Next.js projects | | `mobile-app-builder` | Native iOS/Android + React Native/Flutter specialist | | `solidity-smart-contract-engineer` | EVM contracts, gas optimization, DeFi protocol security | | `sre-site-reliability-engineer` | SLOs, error budgets, chaos engineering, toil reduction | | `embedded-firmware-engineer` | Bare-metal / RTOS for ESP32, STM32, Nordic nRF | | `godot-multiplayer-engineer` | MultiplayerAPI, ENet/WebRTC, RPCs for Godot 4 | | `unity-architect` | ScriptableObjects, data-driven modularity, ECS/DOTS | | `macos-spatial-metal-engineer` | Swift + Metal, high-performance 3D, Vision Pro | | `lsp-index-engineer` | Language Server Protocol, semantic code intelligence | | `+ 116 more` | AI/ML, backend, DevOps, game engines, XR, blockchain... | ### 🎨 Design Personas — 15 agents | Agent | What it does | |---|---| | `design-ui-designer` | Visual design systems, component libraries, brand consistency | | `design-ux-architect` | CSS architecture, accessible layouts, implementation guidance | | `design-brand-guardian` | Brand voice, visual identity enforcement, positioning | | `design-ux-researcher` | Usability testing, behavior analysis, actionable research | | `design-image-prompt-engineer` | Expert prompts for Midjourney, DALL·E, Stable Diffusion | | `design-inclusive-visuals-specialist` | Bias-free, culturally accurate AI imagery | | `+ 9 more` | Motion designers, whimsy injectors, visual storytellers... | ### ⚙️ Operations Personas — 33 agents | Agent | What it does | |---|---| | `code-reviewer` | Constructive review focused on correctness, security, and maintainability | | `react-reviewer` | React-specific patterns, hooks discipline, render performance | | `typescript-reviewer` | Strict type safety, interface contracts, inference quality | | `codebase-onboarding-engineer` | Helps new devs understand unfamiliar repos from the source | | `database-reviewer` | Schema design, index strategy, N+1 detection | | `automation-governance-architect` | Audits automation value, risk, and maintainability | | `+ 27 more` | Go, Python, Java, C#, C++ reviewers + PM roles... | ### 💼 Business Strategy — 14 agents | Agent | What it does | |---|---| | `chief-of-staff` | Filters noise, owns processes, routes decisions for executives | | `executive-brief` | C-suite summaries using McKinsey SCQA + BCG Pyramid | | `feedback-synthesizer` | Turns raw user feedback into quantified product priorities | | `behavioral-nudge-engine` | Behavioral psychology to maximize user motivation | | `government-digital-presales-consultant` | ToG presales, policy interpretation, bid documents | | `cross-border-ecommerce` | Amazon, Shopee, Lazada, AliExpress full-funnel ops | | `+ 8 more` | GTM planners, trend researchers, proposal strategists... | ### 🔬 Specialized — 44 agents Niche experts covering legal tech, academia, finance, spatial computing, and more. | Agent | What it does | |---|---| | `legal-document-review` | Contract review, risk clause flagging, version comparison | | `blockchain-security-auditor` | Smart contract vulnerability detection, audit reports | | `academic-anthropologist` | Culturally coherent society design, kinship systems | | `academic-psychologist` | Psychologically credible characters and interactions | | `workflow-architect` | Maps every workflow path before a line of code is written | | `mcp-builder` | Builds Model Context Protocol servers for AI agents | | `+ 38 more` | Civil engineers, supply chain, HR, real estate, XR, finance... | --- ## 📋 Workflow Checklists — 521 total ### 🛠️ Development — 363 workflows The largest domain. Complete execution checklists covering every phase of software delivery. **Highlights:** `tdd-workflow` · `react-patterns` · `senior-fullstack` · `production-deployment` · `launch-strategy` · `stripe-integration-expert` · `liquid-glass-design` · `motion-advanced` · `nextjs-patterns` · `swift-actor-persistence` · `remote-mcp-server` ### 🔒 Security & Compliance — 64 workflows `owasp-threat-modeling` · `soc2-compliance` · `gdpr-privacy` · `ai-security` · `threat-detection` · `security-review` · `blockchain-audit` · `dependency-auditor` ### 🧪 QA & Testing — 32 workflows `react-testing` · `browser-qa-testing` · `ab-test-setup` · `accessibility` · `ai-regression-testing` · `api-test-suite-builder` ### 📊 Business Ops — 62 workflows `github-ops` · `analytics` · `board-deck-builder` · `autonomous-loops` · `research` · `senior-secops` --- ## 🎨 Rules & Taste Tokens — 41 files The `rules/` directory is the most impactful single folder. It shapes *how* the model generates, not just what it generates. **`rules/taste/`** — Eliminates visual mediocrity: - `animate` — Spring physics, frame-perfect motion, no CSS transitions - `bolder` — Forces decisive type hierarchy, no weak grays - `brand` — HSL-based color theory, custom font stacks - `liquid-glass` — Glassmorphism with correct backdrop-filter values - `motion-advanced` — Bezier curves, staggered reveals, scroll-triggered animations - `overdrive` — Maximum visual density when you need premium impact **`rules/core/`** — Zero-tolerance code hygiene: no magic values, no hardcoded secrets, no silent failures. --- ## ⚙️ IDE Integration ### Cursor ```bash # Install a workflow as .cursorrules ./proagents install react-patterns --cursor # Or reference specific rules in Cursor Settings → Rules for AI cat rules/taste/animate.md ``` ### Claude Code ```bash # Append any persona to CLAUDE.md ./proagents install code-reviewer --stdout >> CLAUDE.md # Load a taste rule for visual work ./proagents install brand --stdout >> CLAUDE.md ``` ### Windsurf / Zed / Trae ```bash # Output to stdout and paste into system instructions ./proagents install ux-architect --stdout ``` ### Gemini CLI ```bash # Install a skill file to ~/.gemini/extensions/ ./proagents install senior-fullstack --target ~/.gemini/extensions/proagents/SKILL.md ``` --- ## 🤝 Contributing 1. Fork the repo 2. Add your `.md` file in the appropriate `personas/` or `workflows/` domain 3. No YAML frontmatter required — clean markdown is enough 4. Open a PR --- ## ⚖️ License & Credits Consolidated from 6 open-source frameworks under MIT and Apache 2.0 licenses. Original attributions preserved in [CREDITS.md](./CREDITS.md). [MIT License](LICENSE) © 2026 Arlandaren
Instalasi
Pasang skill dari https://github.com/Arlandaren/proagentsSkill ini belum dipublikasikan di registri ClawHub. Tempel prompt ke coding agent Anda — Claude Code, Cursor, Codex — dan skill-nya akan dipasang langsung dari repositori sumbernya.
Buka repositori