sunnyeung369/human-first-agent-writing

🎯 A practical multi-agent writing workflow for research, drafting, editing, and fact-checking.

Drawing on NewType-Profile’s role division, complex tasks are completed through research, writing, editing, and verification collaboration. It does not promise to improve automatically but clearly states responsibility and evidence at every step.

Feature Introduction • Quick Start • Usage Examples • Documentation • Contributions

English • Chinese Simplified

---

✨ What problem does this Skill solve?

🤔 Limitations of Single AI:

• Information collection is not comprehensive enough
• Unstable quality of creation
• Lack of professional validation
• Weak knowledge retrieval ability

✅ It offers constraints:

• 6 specialized agents each with their own roles
• Structured workflows Assist with review and improvement
• Multiple Verification Mechanism Ensures accuracy
• Flexible Collaboration Mode Adapts to Various Scenarios

Writing by default follows the Human Voice protocol: specific, restrained, allowing uncertainty, removing clichés and unsubstantiated “professional pitch.”

**First think things through, then write your words like a person. **

---

📊 Core Features

👀 Let’s look at the results first

• 30-second demo: Copy a prompt and observe the complete workflow.
• Before-and-after comparison example: See how it removes AI cavities while retaining judgment and boundaries.
• Feedback template: Submit real usage scenarios, no need for exaggerated ratings.

🎭 6 Major Professions of Agent

Method 1️⃣: Have the editor-in-chief coordinate (recommended for complex tasks)

Chief Auto:

• ✅ Break down the task into 6 stages
• ✅ Coordinate all Agent collaborations
• ✅ Set up quality checkpoints
• ✅ Integrate the final output

Method 2️⃣: Specify a specific Agent (recommended for single tasks)

Method 3️⃣: Task Classification (Quick Mode)

---