Hybrid Model Workflow — AI skill for Claude Code
HAL 分层混合模型工作流 — 强模型(Claude)负责理解/拆解/验收,低成本模型(DeepSeek)负责检索/提取/清洗.
How to install Hybrid Model Workflow
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open ph4ble/hybrid-model-workflow and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Hybrid Model Workflow does
HAL 分层混合模型工作流 — 强模型(Claude)负责理解/拆解/验收,低成本模型(DeepSeek)负责检索/提取/清洗。Hermes Agent skill。
Alternatives in AI
- EchoBird — One-click install + model switch:Claude Code,Codex CLI (OpenAI), Grok Build (xAI), DeepSeek Harness, Kimi Code 3.1k ★
- Loushang — AI-native agent harness for coding workflows by python: multi-model LLM orchestration, stateful sessions, tool 1.3k ★
- Codeseek — Rust-powered code intelligence CLI for AI coding agents 764 ★
README
HAL 分层混合模型工作流

强模型 + 低成本模型分层协作方案,适用于 [Hermes Agent](https://github.com/hermes-ai/hermes-agent) 及任何多模型编排场景。
核心理念
不是所有任务都值得用最贵的模型。这套工作流把任务按复杂度分层:
- 强模型(Claude / GPT-5):负责理解、拆解、判断、最终校验
- 低成本模型(DeepSeek V3.1 / Gemini Flash):负责检索、提取、清洗、摘要
低成本模型只产出中间件,不主导最终结论;强模型对所有交付结果负责。
工作流全景

特性
- ⚡ 智能路由 — 4 条任务路径自动分类(直接执行 / 先拆解 / 强模型主导 / 信息不足)
- 🛡️ 双层验收 — 强模型 5 维度审核 + 8 项 Checklist 全过才交付
- 💰 节省 70%+ — 低成本模型 ¥2/M tokens,强模型仅用于验收收口
- 🔒 零幻觉 — 禁止编造 · 证据优先 · 不确定就明确说不确定
- 🔌 模型无关 — 兼容 Claude、DeepSeek、GPT-5、Gemini、Qwen 等
包含内容
- 📋 四种任务路由规则(Route A-D)
- 📝 低成本执行层 Prompt 模板
- ✅ 强模型 Reviewer Prompt 模板
- 🔍 最终校验 Checklist(8 维度)
- ⚙️ Hermes Agent 配置方法
- 📦 结果回写规则
快速开始
安装为 Hermes Skill
# 克隆到 Hermes skills 目录
mkdir -p ~/.hermes/skills/mlops
git clone https://github.com/ph4ble/hybrid-model-workflow.git ~/.hermes/skills/mlops/hybrid-model-workflow
配置
- 在
~/.hermes/config.yaml添加低成本模型 provider - 可选启用 smart_model_routing
- 详见 SKILL.md 中的配置说明
迭代历史
| 版本 | 重点 |
|---|---|
| V1 | 分层架构,ABCD 任务分类 |
| V2 | 小模型强约束 + 强模型校验 |
| V3 | 可执行工作流规范 |
| V4 | 真实工作流蓝图(路由器、中间件、回写) |
| V6 | 系统提示词 + Hermes 配置 + 运行信息显示 |
License
MIT
Related Skills
Enfold
Enfold. Every memory enfolds the whole. Local, private long-term memory for AI agents: one SQLite brain shared
Nexus Memory
Universal Memory Layer for AI Agents — self-hosted, hybrid search, 3-way: Hermes Plugin · OpenClaw Plugin · MC
MultiAgentDocumentAssistant
A multi-agent research paper assistant with durable LangGraph workflows, three-layer long-term memory, hybrid
Memtrace Public
Structural memory for AI coding agents. Bi-temporal graph, MCP-native, zero LLM calls. Cursor · Claude Code ·
Agent Fix
Universal repair skill & CLI for AI coding agents (Claude Code, Codex, OpenCode, Hermes) — npm postinstall, GU
BladeX
Self-hosted memory-first proxy for AI agents. Cross-session & cross-agent memory + task ledger. Zero code chan
Related Agents
Deepseek Worker
Coding worker on DeepSeek, served by the multi-model gateway's upstream. Use when the user says "use deepseek"
Astro Converter
Specialized agent for autonomous design-to-Astro conversion using a hybrid model. Emits zero-JS static .astro
Chair Advocate
The advocate chair in the docs-assist authoring loop. Shapes a cold packet into a document and declares, as a