FAEM（Felix Agent Evolution Methodology｜Felix Agent 演進方法論） 是由 Felix Hsieh（謝時全） 提出的一套 AI Agent 建構、評估與演進的方法論。 FAEM 的核心理念，是讓 AI Agent 不只是「執行 Prompt」，而是能夠透過結構化的方法進行： 建構 → 評估 → 學習 → 演進 本 Repository 收錄基於 FAEM 方法論所開發的實用型 Agent Skills，將方法論轉化為可以實際使用、重複執行與持續擴充的 AI Agent 能力。
FAEM (Felix Agent Evolution Methodology) is a methodology created by Felix Hsieh for building, evaluating, and evolving AI Agents. The core idea of FAEM is to move AI Agents beyond simply executing prompts toward structured capabilities that can be: Built → Evaluated → Learned from → Evolved This repository contains practical Agent Skills developed based on the FAEM methodology, transforming its principles into reusable, executable, and extensible AI Agent capabilities.
FAEM 專注於解決企業導入 AI Agent 時的一個核心問題：
如何讓 AI Agent 從「會回答問題」，進一步演進成「具備專業能力、可以被評估，而且能持續改善的 Agent」。 FAEM 不只關注 Prompt Engineering，而是從完整的 Agent Lifecycle 來思考 AI 系統：
- Build｜建構 — 定義 Agent 的角色、能力、知識與工作流程
- Evaluate｜評估 — 評估輸出品質、專業程度與能力覆蓋
- Learn｜學習 — 從使用結果與評估結果中找出不足
- Evolve｜演進 — 持續改善 Agent 的能力與架構 因此，FAEM 的目標不是建立一次性的 Prompt，而是建立可以持續演進的 AI Agent。
FAEM focuses on a fundamental challenge in enterprise AI adoption:
How can an AI Agent evolve from simply answering questions into an Agent with professional capabilities that can be evaluated and continuously improved? FAEM goes beyond Prompt Engineering and approaches AI systems through a complete Agent Lifecycle:
- Build — Define roles, capabilities, knowledge, and workflows
- Evaluate — Assess output quality, expertise, and capability coverage
- Learn — Identify weaknesses from usage and evaluation results
- Evolve — Continuously improve the Agent's capabilities and architecture The goal of FAEM is therefore not to create one-time prompts, but to build AI Agents that can continuously evolve.
FAEM 的核心循環為：
Build
↓
Evaluate
↓
Learn
↓
Evolve
↺
也就是： 建構 → 評估 → 學習 → 演進 → 再次建構 這個循環讓 Agent 能夠從實際使用與評估結果中持續獲得改善，而不是停留在一次性的 Prompt 設計。
The core FAEM lifecycle is:
Build
↓
Evaluate
↓
Learn
↓
Evolve
↺
This creates a continuous improvement loop in which Agents learn from real-world usage and evaluation results rather than remaining static, one-time prompt configurations.
FAEM Prompt Auditor 是第一個基於 FAEM 方法論所開發的 Agent Skill。 它可以對 Prompt 進行系統化 Audit，分析一個 Prompt 背後需要哪些專家角色、領域知識與專業能力，並評估目前 Prompt 所涵蓋的專業程度。 主要能力包括：
- Expert Role Detection｜專家角色辨識
- Professional Coverage｜專業能力覆蓋度評估
- Dual Score Model｜雙維度評分模型
- Agent Evolution｜Agent 演進分析
- Structured Audit Report｜結構化 Audit 報告 適用於：
- Prompt Engineering
- AI Agent Development
- Enterprise AI
- AI Workflow Optimization
- Agent Capability Assessment
- AI 專業能力分析
詳細內容請參閱：
faem-prompt-auditor/SKILL.md
FAEM Prompt Auditor is the first Agent Skill developed based on the FAEM methodology. It systematically audits prompts by identifying the expert roles, domain knowledge, and professional capabilities required to produce high-quality results, while evaluating the level of professional coverage provided by the prompt. Key capabilities include:
- Expert Role Detection
- Professional Coverage Assessment
- Dual Score Model
- Agent Evolution Analysis
- Structured Audit Reports It can be applied to:
- Prompt Engineering
- AI Agent Development
- Enterprise AI
- AI Workflow Optimization
- Agent Capability Assessment
- Professional Capability Analysis
For detailed information, see:
faem-prompt-auditor/SKILL.md
目前 Repository 結構如下：
faem-skills/
│
├── faem-prompt-auditor/
│ ├── SKILL.md
│ ├── README.md
│ ├── references/
│ ├── scripts/
│ └── templates/
│
└── README.md
未來將持續加入更多基於 FAEM 方法論所開發的 Agent Skills。
The current repository structure is:
faem-skills/
│
├── faem-prompt-auditor/
│ ├── SKILL.md
│ ├── README.md
│ ├── references/
│ ├── scripts/
│ └── templates/
│
└── README.md
Felix Hsieh（謝時全） FAEM — Felix Agent Evolution Methodology 創建者。 FAEM 是一套持續發展中的 AI Agent 建構與演進方法論，旨在探索如何讓 AI 從單純的 Prompt / Chatbot，逐步演進成具備專業能力、可評估、可改善的 Agent System。
Felix Hsieh Creator of FAEM — Felix Agent Evolution Methodology. FAEM is an evolving methodology for building and improving AI Agents, exploring how AI can progress from simple prompts and chatbots toward professional, evaluable, and continuously improving Agent Systems.
From Prompt Engineering to Agent Evolution.
從 Prompt Engineering，走向 Agent Evolution。 FAEM 的長期目標，是建立一套可以協助企業與 AI 開發者： 建構、評估與持續演進 AI Agent 的方法論與工具生態。
From Prompt Engineering to Agent Evolution. The long-term vision of FAEM is to establish a methodology and ecosystem of tools that helps enterprises and AI developers: Build, evaluate, and continuously evolve AI Agents.
See the individual Skill directory for licensing information.