AI To Agent Custom Skill — AI skill for Claude Code
AI to Agent 上位決策治理 Skill — Context over Dogma, Problems over Benchmarks, People over Models, Outcome over Complexity.
How to install AI To Agent Custom Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open draiagent/ai-to-agent-custom-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What AI To Agent Custom Skill does
AI to Agent 上位決策治理 Skill — Context over Dogma, Problems over Benchmarks, People over Models, Outcome over Complexity.
Alternatives in AI
- Define — Definition phase - Clarify and scope problems with multi-AI consensus 2.8k ★
- Agentmemory — #1 Persistent memory for AI coding agents based on real-world benchmarks 1.9k ★
- Gptaku Plugins — A Claude Code plugin marketplace for people who want to become AI Native 1.1k ★
README
Context-First Human Agent
**AI to Agent 上位決策治理 Skill**
**Context over Dogma · Problems over Benchmarks · People over Models · Outcome over Complexity**
核心主張
世界不存在一條適用所有人的標準道路。
Agent 應先理解**人、任務、情境、限制、時間與風險**,再決定該使用 Rule、Code、Search、RAG、LLM、Agent 或 Human-in-the-loop。
這個 Skill 的目的不是讓 Agent 更會「展示聰明」,而是讓它更會:
- 找出真正問題。
- 選擇最小充分複雜度。
- 根據使用者與任務動態調整。
- 避免 Benchmark Worship、Model Worship、Agent Everywhere。
- 將 AI 能力轉換成可理解、可執行、可驗證的人類成果。
四大原則
- Context over Dogma|情境高於教條
- Problems over Benchmarks|問題高於跑分
- People over Models|人高於模型
- Outcome over Complexity|結果高於複雜度
核心哲學
**AI 的目的,不是證明機器有多聰明,而是讓更多人有能力把事情做好。**
視覺圖卡
01 / 04|讓 AI 回歸人的真實問題

**People over Models。**
AI 的價值不在於模型多聰明,而在於能不能讓更多人:
- 看得懂
- 用得起
- 做得到
中心的「人的需求」代表真正的最佳化目標是 **Human Outcome**。
02 / 04|先理解人,再決定方法

**Context over Dogma。**
先判斷:
- 誰使用
- 做什麼
- 什麼環境
- 有何限制
- 目前階段
原則保持穩定,方法因境調整。
03 / 04|選擇足夠好用的工具

**Minimum Sufficient Intelligence。**
- 固定規則 → Rule
- 數值計算 → Code
- 即時資訊 → Search
- 複雜推理 → LLM
- 跨工具執行 → Agent
- 高風險決策 → Human-in-the-loop
圖卡是教學收斂版;完整路由另包含 RAG、Structured Processing、Choice、Score、Extract。
04 / 04|把人的需求,變成可驗證成果

**VAD / VAC → Execution → Verification。**
理解需求
↓
VAD:讓人看懂任務
↓
VAC:形成可執行、可驗證的任務契約
↓
執行工具
↓
驗證成果
同時遵守:
History ≠ State
Events → Semantic State → Relevant Context → Decision
最後驗收:
- 問題解決了嗎?
- 使用者能行動嗎?
- 成本與風險合理嗎?
一致性檢查
四張圖卡與 `SKILL.md` 核心邏輯一致:
People
↓
Context
↓
Routing
↓
VAD / VAC
↓
Execution
↓
Verified Outcome
已統一的正式用語
- 「複雜理解 → 模型」在正式文件中統一為 「複雜推理 → LLM」。
- 「VAD 看懂任務」對應正式定義:人類可理解的任務視覺化。
- 「VAC 說清規格」對應正式定義:機器可執行、可驗證的任務契約。
- 圖卡是教學精簡版;完整 Task Routing 以
SKILL.md為準。
與 AI to Agent 的關係
Human Need
↓
Context
↓
Related Skills
Codenano
Lightweight AI coding agent SDK. Same power, 97% less complexity. Build custom AI coding agents in 1 minutes
Brick SR1
brick is a smart AI Models router, based on complexity & capabilities extraction from the query to the models
9router Auto Free
🚀 Automated scraper & injector for 100+ free AI coding models into 9router combos. Live pre-tested, ranked by
OmniGlyph
Cut your Claude bill 59–70% by rendering bulky LLM context as dense PNG pages — 100% read accuracy, exact per-
Comfyui LLM Prompt Enhancer
A ComfyUI custom node that uses GPT, Claude, Gemini, or local LLMs to enhance your rompts with 50+ artistic st
Patch Cc
Interactive patcher for the Claude Code native binary. Live inline thinking, detailed tool calls, subagent mod
Related Agents
Selfish Audience Intelligence
Build an evidence-based model of the people the user can help, the problems they experience, and the language/
WFL
Researches new ComfyUI models, workflows, and custom nodes, and reports which ones are worth adding to OpenLay
AI Data Scientist
Data Scientist who applies statistical analysis, machine learning, and AI techniques to extract insights, buil