#subagent #routing #multi-provider #task-management

Hive Task Router

Intelligent task routing system that identifies task types (web/code/data/doc/chat) and routes to optimal models with appropriate execution mode (subagent/main session). Supports any AI model provider via environment variables.

qiongcao@qiongqiongcaocao

Install

openclaw skills install @qiongqiongcaocao/hive-task-router

Hive Task Router

Smart task distribution system for OpenClaw — Automatically routes tasks to optimal models and execution modes based on task type recognition.

Universal Model Support — Works with any AI model provider (Bailian, OpenAI, Anthropic, etc.) via environment variable configuration.

When to Use

Automatic Trigger Rules (For AI Agents)

Use this skill automatically when the user's message contains ANY of these patterns:

Trigger Pattern	Keywords	Action
Code tasks	代码、编程、脚本、函数、nodejs、react、vue、typescript、javascript、html、css、前端、后端、api、接口、调试、bug、优化、重构	Run `router.sh` → Execute with `code` model + subagent
Web research	搜索、查找、调研、研究、github、项目、趋势、报告、分析、对比、评测、最新、2026、新闻、动态	Run `router.sh` → Execute with `web` model + subagent
Data tasks	数据、分析、统计、图表、可视化、excel、csv、json、处理、转换	Run `router.sh` → Execute with `data` model + subagent
Documentation	文档、说明、教程、指南、手册、readme、wiki、注释、文档化	Run `router.sh` → Execute with `doc` model + subagent
Batch tasks	多个、批量、同时、并行、一起	Run `router.sh` for each → Execute in parallel

Do NOT use this skill when:

Simple chat/conversation (你好、谢谢、安排、计划、汇报、总结)
Quick questions (what is, how to, explain)
Tasks already in progress
User explicitly says "don't use subagent" or "just answer directly"

Manual Trigger (For Users)

Users can explicitly trigger this skill by:

Running router.sh "task description" directly
Saying "use hive router" or "analyze this task"
Asking "which model should I use for this task"

Decision Flow for AI Agents

text

Receive user message
    ↓
Contains specific trigger keywords? (code/web/data/doc/batch)
    ↓
YES → Run router.sh to analyze
    ↓
Get recommended model + execution mode
    ↓
Execute with recommended configuration
    ↓
Report result to user
    ↓
NO → Check for vague task keywords? (任务、帮忙、处理、搞定、完成)
    ↓
YES → Ask clarifying question (see "Vague Task Handling")
    ↓
User clarifies → Re-analyze with new info
    ↓
NO → Handle directly (no routing needed)

Vague Task Handling

When user message is vague (e.g., "做个任务", "帮忙处理一下", "搞定这件事"):

Step 1: Acknowledge and ask

text

好的主人，请问是什么类型的任务？

💻 写代码/脚本？
   - 例如："写个 Python 脚本"、"开发一个 API"

🔍 搜索调研？
   - 例如："搜索最新趋势"、"调研竞品"

📊 数据处理？
   - 例如："分析 Excel 数据"、"转换 JSON 格式"

📄 写文档？
   - 例如："写 API 文档"、"编写教程"

💬 还是只是聊天？
   - 例如："今天有什么安排"、"帮我总结一下"

或者您直接告诉我具体内容，我来判断！

Step 2: User clarifies

text

User: "写个脚本处理数据"
  ↓
Now contains: "脚本" (code) + "数据" (data)
  ↓
Priority: code > data
  ↓
Execute with: qwen3-coder-plus + subagent

Vague Keywords (trigger clarification):

任务、帮忙、处理、搞定、完成、做一下、弄一下
这个、那个、一件事、一个东西

Specific Keywords (trigger automatic routing):

code: 脚本、代码、编程、函数、nodejs、react...
web: 搜索、调研、分析、趋势、报告、对比...
data: 数据、统计、图表、excel、csv、json...
doc: 文档、教程、指南、readme、wiki...

Ideal scenarios:

Handling multiple concurrent tasks
Technical development requiring code-specialized models
Research tasks needing web search capabilities
Mixed workloads with varying complexity
Multi-provider environments (Bailian + OpenAI + Anthropic)

Quick Reference

Task Type Recognition Keywords

Type	Keywords (Chinese)	Keywords (English)	Priority
web 🔍	搜索、查找、调研、研究、github、项目、趋势、报告、分析、对比、评测、最新、2026、新闻、动态	search, research, github, project, trend, report, analysis, comparison, latest, news	1 (Highest)
code 💻	代码、编程、脚本、函数、nodejs、react、vue、typescript、javascript、html、css、前端、后端、api、接口、调试、bug、优化、重构	code, programming, script, function, nodejs, react, vue, typescript, javascript, html, css, frontend, backend, api, debug, bug, optimize, refactor	2
data 📊	数据、分析、统计、图表、可视化、excel、csv、json、处理、转换	data, analysis, statistics, chart, visualization, excel, csv, json, processing, conversion	3
doc 📄	文档、说明、教程、指南、手册、readme、wiki、注释、文档化	documentation, guide, tutorial, manual, readme, wiki, comment, document	4
chat 💬	你好、谢谢、再见、今天、明天、安排、计划、汇报、总结、提醒、备忘	hello, thanks, goodbye, today, tomorrow, plan, schedule, summary, reminder, memo	5 (Default)

Model Selection Rules

Note: Model IDs are configurable via environment variables. Replace provider/ with your actual model provider (e.g., bailian/, openai/, anthropic/).

Task Type	Default Model	Environment Variable	Reason
code	`provider/qwen3-coder-plus`	`HIVE_MODEL_CODE`	Specialized in code generation and debugging
web	`provider/qwen3-max`	`HIVE_MODEL_WEB`	Strong search and reasoning capabilities
data	`provider/qwen3-coder-plus`	`HIVE_MODEL_DATA`	Code-based data processing
doc	`provider/qwen3.5-plus`	`HIVE_MODEL_DOC`	Good text generation, cost-effective
chat	`provider/qwen3.5-plus`	`HIVE_MODEL_CHAT`	Best for casual conversation, cost-effective

Model Configuration Examples

Bailian (通义千问):

bash

export HIVE_MODEL_CODE="bailian/qwen3-coder-plus"
export HIVE_MODEL_WEB="bailian/qwen3-max-2026-01-23"
export HIVE_MODEL_CHAT="bailian/qwen3.5-plus"
export HIVE_MODEL_DOC="bailian/qwen3.5-plus"
export HIVE_MODEL_DATA="bailian/qwen3-coder-plus"

Model Validation Modes

Automatic Model Detection (Recommended):

bash

# Auto-detect available models from OpenClaw
export HIVE_VALIDATE_MODEL=auto

First run: Detects models and caches configuration
Subsequent runs: Uses cached config (24h TTL)
Benefit: No manual configuration needed!

Manual Validation Modes:

Mode	Environment Variable	Behavior	Use Case
Auto (Recommended)	`export HIVE_VALIDATE_MODEL=auto`	Auto-detect + cache 24h	Best for most users
Cache	`export HIVE_VALIDATE_MODEL=cache`	Validate once, cache 24h	Manual config, stable
Always	`export HIVE_VALIDATE_MODEL=1`	Validate every execution	Debugging, changes
Never	`export HIVE_VALIDATE_MODEL=0`	Skip validation	Production, known config

Cache Configuration:

bash

# Cache directory (default: ~/.hive-task-router)
export HIVE_CACHE_DIR="$HOME/.hive-task-router"

# Cache TTL in seconds (default: 86400 = 24 hours)
export HIVE_CACHE_TTL=86400

Validation Behavior:

✅ Checks if model IDs contain provider/ placeholder
✅ Warns if placeholder detected
✅ Caches validation result (configurable TTL)
✅ Non-blocking (warnings only, doesn't stop execution)

OpenAI:

bash

export HIVE_MODEL_CODE="openai/gpt-4"
export HIVE_MODEL_WEB="openai/gpt-4-turbo"
export HIVE_MODEL_CHAT="openai/gpt-3.5-turbo"
export HIVE_MODEL_DOC="openai/gpt-3.5-turbo"
export HIVE_MODEL_DATA="openai/gpt-4"

Anthropic (Claude):

bash

export HIVE_MODEL_CODE="anthropic/claude-3-5-sonnet"
export HIVE_MODEL_WEB="anthropic/claude-3-opus"
export HIVE_MODEL_CHAT="anthropic/claude-3-haiku"
export HIVE_MODEL_DOC="anthropic/claude-3-haiku"
export HIVE_MODEL_DATA="anthropic/claude-3-5-sonnet"

Mixed Providers:

bash

# Use best model for each task type
export HIVE_MODEL_CODE="anthropic/claude-3-5-sonnet"  # Best for code
export HIVE_MODEL_WEB="openai/gpt-4-turbo"           # Best for search
export HIVE_MODEL_CHAT="bailian/qwen3.5-plus"        # Cost-effective

Execution Mode Rules

Task Type	Execution Mode	Reason
chat	Main Session	Quick response, no need for isolation
code/web/data/doc	Subagent	Long-running tasks, parallel execution, session isolation

Priority Rule: When multiple keywords match, use the highest priority type (web > code > data > doc > chat).

Usage

For AI Agents (Automatic Integration)

When installed as an OpenClaw Skill, the agent will automatically use this skill when:

User message contains task keywords (code/web/data/doc/batch)
Task is long-running (not a simple Q&A)
Multiple tasks need parallel execution

Agent Decision Example:

text

User: "帮我写一个 Python 脚本处理 Excel 数据"
  ↓
Agent checks: Contains "脚本" (code) + "数据" (data)
  ↓
Priority: code > data
  ↓
Agent executes:
openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "帮我写一个 Python 脚本处理 Excel 数据"

Configuration for Agents: Add to agent's AGENTS.md or SOUL.md:

markdown

## Hive Task Router Integration

When receiving tasks:
1. Check if message contains task keywords (see SKILL.md)
2. If yes → Use hive-task-router skill
3. If no → Handle directly

Method 1: Router Script (Recommended)

The router script automatically analyzes tasks and outputs recommended execution commands.

bash

# Basic usage
bash router.sh "帮我写一个 Node.js 脚本"

# Analyze research task
bash router.sh "搜索 2026 年最新的前端趋势"

# Analyze data task
bash router.sh "分析这个 JSON 数据并生成图表"

Output format:

text

================================
蜂巢智能任务分发系统 - 路由分析
================================

任务描述：帮我写一个 Node.js 脚本
任务类型：code
推荐模型：bailian/qwen3-coder-plus
执行方式：subagent

📦 代码任务 - 使用 qwen3-coder-plus 模型
   适合：Node.js、前端代码、脚本编写

================================
推荐执行命令:
================================

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "帮我写一个 Node.js 脚本"

Method 2: Manual Commands

Code Tasks

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "帮我写一个 Express API 服务"

Web Research Tasks

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-max-2026-01-23 \
  --task "调研 5 个 React UI 库"

Chat Tasks (Main Session)

bash

openclaw agent \
  --session-id agent:main:chat \
  --model bailian/qwen3.5-plus \
  --message "今天有什么安排"

Data Processing Tasks

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "处理这个 CSV 文件并生成统计报告"

Documentation Tasks

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3.5-plus \
  --task "为这个项目编写 README 文档"

Method 3: Parallel Execution

For batch tasks, use parallel subagents:

bash

# Spawn multiple subagents concurrently
openclaw sessions spawn --mode run --runtime subagent --model bailian/qwen3-max-2026-01-23 --task "调研项目 A" &
openclaw sessions spawn --mode run --runtime subagent --model bailian/qwen3-max-2026-01-23 --task "调研项目 B" &
openclaw sessions spawn --mode run --runtime subagent --model bailian/qwen3-max-2026-01-23 --task "调研项目 C" &

# Wait for all to complete
wait

# Then collect and summarize results

Examples

Example 1: Code Development Task

User Input:

text

帮我写一个 Node.js 文件处理脚本，支持读取 CSV 和 JSON 格式

Router Analysis:

Matched Keywords: Node.js, 脚本，文件处理
Task Type: code
Recommended Model: bailian/qwen3-coder-plus
Execution Mode: subagent

Execution Command:

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "帮我写一个 Node.js 文件处理脚本，支持读取 CSV 和 JSON 格式"

Example 2: Web Research Task

User Input:

text

搜索 2026 年最新的前端趋势，包括 React、Vue、Svelte 的对比

Router Analysis:

Matched Keywords: 搜索，2026, 趋势，对比
Task Type: web (Priority 1)
Recommended Model: bailian/qwen3-max-2026-01-23
Execution Mode: subagent

Execution Command:

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-max-2026-01-23 \
  --task "搜索 2026 年最新的前端趋势，包括 React、Vue、Svelte 的对比"

Example 3: Daily Chat Task

User Input:

text

今天有什么安排？帮我总结一下昨天的工作

Router Analysis:

Matched Keywords: 今天，安排，总结
Task Type: chat
Recommended Model: bailian/qwen3.5-plus
Execution Mode: main_session

Execution Command:

bash

openclaw agent \
  --session-id agent:main:chat \
  --model bailian/qwen3.5-plus \
  --message "今天有什么安排？帮我总结一下昨天的工作"

Example 4: Data Analysis Task

User Input:

text

分析这个销售数据 Excel 文件，生成可视化图表和统计报告

Router Analysis:

Matched Keywords: 分析，数据，Excel, 图表，统计
Task Type: data
Recommended Model: bailian/qwen3-coder-plus
Execution Mode: subagent

Execution Command:

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3-coder-plus \
  --task "分析这个销售数据 Excel 文件，生成可视化图表和统计报告"

Example 5: Documentation Task

User Input:

text

为这个 Python 项目编写完整的 API 文档和使用教程

Router Analysis:

Matched Keywords: 文档，教程
Task Type: doc
Recommended Model: bailian/qwen3.5-plus
Execution Mode: subagent

Execution Command:

bash

openclaw sessions spawn \
  --mode run \
  --runtime subagent \
  --model bailian/qwen3.5-plus \
  --task "为这个 Python 项目编写完整的 API 文档和使用教程"

Example 6: Mixed Task Batch (Parallel)

User Input (Multiple Tasks):

text

1. 写个脚本处理 JSON 数据
2. 搜索最新 AI 工具
3. 今天有什么安排

Router Analysis:

Task 1: code → qwen3-coder-plus + subagent
Task 2: web → qwen3-max-2026-01-23 + subagent
Task 3: chat → qwen3.5-plus + main_session

Parallel Execution:

bash

# Task 1 & 2 run in parallel subagents
openclaw sessions spawn --mode run --runtime subagent --model bailian/qwen3-coder-plus --task "写个脚本处理 JSON 数据" &
openclaw sessions spawn --mode run --runtime subagent --model bailian/qwen3-max-2026-01-23 --task "搜索最新 AI 工具" &

# Task 3 runs in main session (non-blocking)
openclaw agent --session-id agent:main:chat --model bailian/qwen3.5-plus --message "今天有什么安排"

# Wait for subagents
wait

Performance: 3x faster than sequential execution

Installation

Install via ClawHub

bash

clawhub install qiongcao/hive-task-router

Manual Installation

Clone or download this skill folder

Copy to your OpenClaw skills directory:

bash

cp -r hive-task-router ~/.openclaw/workspace/skills/

Make router script executable:

bash

chmod +x ~/.openclaw/workspace/skills/hive-task-router/router.sh

Configure models for your provider:

bash

export HIVE_MODEL_CODE="bailian/qwen3-coder-plus"
export HIVE_MODEL_WEB="bailian/qwen3-max-2026-01-23"
export HIVE_MODEL_CHAT="bailian/qwen3.5-plus"

Configuration

Prerequisites

OpenClaw CLI installed
Bash shell available
Models configured (adjust for your provider):
- Code model (e.g., bailian/qwen3-coder-plus)
- Web model (e.g., bailian/qwen3-max-2026-01-23)
- Chat model (e.g., bailian/qwen3.5-plus)

Environment Variables

Optional environment variables for customization:

bash

# Model overrides (required)
export HIVE_MODEL_CODE="bailian/qwen3-coder-plus"
export HIVE_MODEL_WEB="bailian/qwen3-max-2026-01-23"
export HIVE_MODEL_CHAT="bailian/qwen3.5-plus"
export HIVE_MODEL_DOC="bailian/qwen3.5-plus"
export HIVE_MODEL_DATA="bailian/qwen3-coder-plus"

# Optional: custom session IDs
export HIVE_SESSION_CODE="custom:code:session"
export HIVE_SESSION_WEB="custom:web:session"
export HIVE_SESSION_CHAT="custom:chat:session"

# Optional: concurrency limit
export HIVE_MAX_CONCURRENT=10

Verify Setup

bash

# Check models
openclaw models list | grep bailian

# Test router script
bash router.sh "测试任务"

# Verify environment variables
echo $HIVE_MODEL_CODE
echo $HIVE_MODEL_WEB

Troubleshooting

Issue 1: Router script not found

bash

# Make sure script is executable
chmod +x router.sh

# Run with full path
bash /path/to/router.sh "task"

Issue 2: Model not available

bash

# Check available models
openclaw models list

# Update environment variables with available models
export HIVE_MODEL_CODE="bailian/qwen3-coder-plus"

Issue 3: Task type misidentified

bash

# Add more specific keywords to router.sh
# Edit CODE_KEYWORDS, WEB_KEYWORDS, etc.

Issue 4: Wrong model used

bash

# Verify environment variables are set
echo $HIVE_MODEL_CODE
echo $HIVE_MODEL_WEB

# Set them explicitly before running router.sh
export HIVE_MODEL_CODE="bailian/qwen3-coder-plus"
bash router.sh "task"

Best Practices

Task Distribution Principles
- ✅ Short tasks → Main session
- ✅ Long tasks → Subagent
- ✅ Batch tasks → Multiple parallel subagents
- ✅ Fixed types → Session isolation + specialized models
Model Selection
- Use code-specialized models for programming tasks
- Use reasoning models for research tasks
- Use cost-effective models for chat tasks
- Mix providers for best-in-class results
Concurrency Control
- Recommended max concurrent subagents: 5-10
- For 10+ tasks → Execute in batches
- Monitor API quota usage
Environment Management
- Set environment variables in .bashrc or .zshrc
- Use different configs for different projects
- Document your model choices

Performance

Metric	Traditional	Hive Router	Improvement
3 project research	~180s	~60s	3x ⚡
Model utilization	Single model	Multi-model	Flexible
Task routing	Manual	Automatic	Intelligent
Multi-provider	Manual switching	Auto config	Seamless

Provider Compatibility

Provider	Status	Notes
Bailian (通义千问)	✅ Tested	Default configuration
OpenAI (GPT)	✅ Compatible	Set HIVE_MODEL_* variables
Anthropic (Claude)	✅ Compatible	Set HIVE_MODEL_* variables
Google (Gemini)	✅ Compatible	Set HIVE_MODEL_* variables
Other OpenAI-compatible	✅ Compatible	Use provider/ prefix

License

MIT License - Feel free to use and modify.

Author: qiongcao
Version: 1.0.0
Last Updated: 2026-03-12
Universal Model Support: Yes

Hive Task Router

Install

Hive Task Router

When to Use

Automatic Trigger Rules (For AI Agents)

Manual Trigger (For Users)

Decision Flow for AI Agents

Vague Task Handling

Quick Reference

Task Type Recognition Keywords

Model Selection Rules

Model Configuration Examples

Model Validation Modes

Execution Mode Rules

Usage

For AI Agents (Automatic Integration)

Method 1: Router Script (Recommended)

Method 2: Manual Commands

Code Tasks

Web Research Tasks

Chat Tasks (Main Session)

Data Processing Tasks

Documentation Tasks

Method 3: Parallel Execution

Examples

Example 1: Code Development Task

Example 2: Web Research Task

Example 3: Daily Chat Task

Example 4: Data Analysis Task

Example 5: Documentation Task

Example 6: Mixed Task Batch (Parallel)

Installation

Install via ClawHub

Manual Installation

Configuration

Prerequisites

Environment Variables

Verify Setup

Troubleshooting

Issue 1: Router script not found

Issue 2: Model not available

Issue 3: Task type misidentified

Issue 4: Wrong model used

Best Practices

Performance

Provider Compatibility

License

Related skills