Skip to content
Agentmart
Submit a tool

deerflow-skill

DeerFlow agent orchestration for complex tasks requiring multi-step reasoning, web search, tool orchestration, or parallel subagent delegation.

Skill by lgwanai

IndexedMIT★ 4
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/lgwanai-deerflow-skill

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

DeerFlow agent orchestration for complex tasks requiring multi-step reasoning, web search, tool orchestration, or parallel subagent delegation. Use when user needs to research topics, search web, or delegate complex tasks to subagents. 使用此 Skill 进行复杂任务的编排和执行。

Use cases: research, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found
Runs as
Python package

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/deerflow-skill

What's inside

Skill name
deerflow-skill
Description
DeerFlow agent orchestration for complex tasks requiring multi-step reasoning, web search, tool orchestration, or parallel subagent delegation. Use when user needs to research topics, search web, or delegate complex tasks to subagents. 使用此 Skill 进行复杂任务的编排和执行。
License
MIT
Instructions
126 lines in SKILL.md · view on GitHub

Before installing

  • Ships executable scripts

Agent Skills spec issues

  • name: "deerflow-skill" does not match folder "deerflow"
  • allowed-tools: must be a space-separated string

Files (190)

  • .gitignore
  • README.md
  • SKILL.md
  • adapter.py
  • config.example.yaml
  • __init__.py · deerflow/
  • __init__.py · deerflow/agents/
  • __init__.py · deerflow/agents/checkpointer/
  • async_provider.py · deerflow/agents/checkpointer/
  • provider.py · deerflow/agents/checkpointer/
  • factory.py · deerflow/agents/
  • features.py · deerflow/agents/
  • __init__.py · deerflow/agents/lead_agent/
  • agent.py · deerflow/agents/lead_agent/
  • prompt.py · deerflow/agents/lead_agent/
  • __init__.py · deerflow/agents/memory/
  • message_processing.py · deerflow/agents/memory/
  • prompt.py · deerflow/agents/memory/
  • queue.py · deerflow/agents/memory/
  • storage.py · deerflow/agents/memory/
  • summarization_hook.py · deerflow/agents/memory/
  • updater.py · deerflow/agents/memory/
  • __init__.py · deerflow/agents/middlewares/
  • clarification_middleware.py · deerflow/agents/middlewares/
  • dangling_tool_call_middleware.py · deerflow/agents/middlewares/
  • deferred_tool_filter_middleware.py · deerflow/agents/middlewares/
  • dynamic_context_middleware.py · deerflow/agents/middlewares/
  • llm_error_handling_middleware.py · deerflow/agents/middlewares/
  • loop_detection_middleware.py · deerflow/agents/middlewares/
  • memory_middleware.py · deerflow/agents/middlewares/
  • sandbox_audit_middleware.py · deerflow/agents/middlewares/
  • subagent_limit_middleware.py · deerflow/agents/middlewares/
  • summarization_middleware.py · deerflow/agents/middlewares/
  • thread_data_middleware.py · deerflow/agents/middlewares/
  • title_middleware.py · deerflow/agents/middlewares/
  • todo_middleware.py · deerflow/agents/middlewares/
  • tool_call_metadata.py · deerflow/agents/middlewares/
  • tool_error_handling_middleware.py · deerflow/agents/middlewares/
  • uploads_middleware.py · deerflow/agents/middlewares/
  • view_image_middleware.py · deerflow/agents/middlewares/
  • thread_state.py · deerflow/agents/
  • client.py · deerflow/
  • __init__.py · deerflow/community/aio_sandbox/
  • aio_sandbox.py · deerflow/community/aio_sandbox/
  • aio_sandbox_provider.py · deerflow/community/aio_sandbox/
  • backend.py · deerflow/community/aio_sandbox/
  • local_backend.py · deerflow/community/aio_sandbox/
  • remote_backend.py · deerflow/community/aio_sandbox/
  • sandbox_info.py · deerflow/community/aio_sandbox/
  • __init__.py · deerflow/community/ddg_search/
  • tools.py · deerflow/community/ddg_search/
  • tools.py · deerflow/community/exa/
  • tools.py · deerflow/community/firecrawl/
  • __init__.py · deerflow/community/image_search/
  • tools.py · deerflow/community/image_search/
  • infoquest_client.py · deerflow/community/infoquest/
  • tools.py · deerflow/community/infoquest/
  • jina_client.py · deerflow/community/jina_ai/
  • tools.py · deerflow/community/jina_ai/
  • tools.py · deerflow/community/tavily/
  • …and 130 more

SKILL.md (first lines)

# DeerFlow Agent

Invoke the DeerFlow agent system directly within Claude Code. No server required - runs embedded in the current process.

## Command System

> **双模式支持**: 本 Skill 同时支持**显性指令**和**模糊匹配**。当用户输入以 `/` 开头的指令时,直接执行对应操作;否则进行意图推断。

### 显性指令列表

| 指令 | 别名 | 功能 | 示例 |
|------|------|------|------|
| `/deer` | `/d`, `/启动` | 启动 DeerFlow agent | `/deer 研究量子计算最新进展` |
| `/deer --flash` | `/d -f`, `/快问` | 快速模式(无思考、无规划) | `/deer --flash 法国首都` |
| `/deer --pro` | `/d -p`, `/专业` | 专业模式(思考+规划) | `/deer --pro 创建REST API项目计划` |
| `/deer --ultra` | `/d -u`, `/超级` | 超级模式(思考+规划+并行子agent) | `/deer --ultra 分析所有模块性能瓶颈` |

### 指令详细说明

#### `/deer` - 启动 Agent
```
/deer <任务描述>
/d <任务描述>  # 别名
/启动 <任务描述>  # 中文别名

示例:
  /deer 研究量子计算的最新进展
  /deer 分析宝马中国市场策略
  /d 对比 Tesla 和 BYD 的技术路线
```

#### `/deer --flash` - 快速模式
```
/deer --flash <简单问题>
/d -f <简单问题>  # 别名
/快问 <简单问题>  # 中文别名

特点: 无思考、无规划、无子agent,快速响应

示例:
  /deer --flash 法国的首都是哪里
  /d -f 今天天气如何
  /快问 1+1等于几
```

#### `/deer --pro` - 专业模式
```
/deer --pro <复杂任务>
/d -p <复杂任务>  # 别名
/专业 <复杂任务>  # 中文别名

特点: 有思考、有规划、无子agent,结构化处理

示例:
  /deer --pro 创建一个REST API的详细项目计划
  /d -p 设计微服务架构方案
  /专业 制定代码重构策略
```

#### `/deer --ultra` - 超级模式
```
/deer --ultra <大型任务>
/d -u <大型任务>  # 别名
/超级 <大型任务>  # 中文别名

特点: 有思考、有规划、有并行子agent,处理复杂任务

示例:
  /deer --ultra 分析所有模块性能并找出瓶颈
  /d -u 对比三家竞品公司的技术栈
  /超级 研究行业发展趋势并生成报告
```

### 模糊匹配关键词

当用户输入不是显性指令时,通过关键词推断意图:

| 意图 | 触发关键词 | 执行模式 |
|------|-----------|----------|
| 简单查询 | 是什么、什么是、多少、哪里 | --flash |
| 网络搜索 | 搜索、查一下、调研、research、search | 标准模式 |
| 复杂任务 | 分析、对比、规划、设计、创建 | --pro |
| 大型任务 | 全面、所有、多个、并行、批量 | --ultra |

## Mode Presets

| Mode | Thinking | Planning | Subagents | Use Case |
|------|----------|----------|-----------|----------|
| `--flash` | No | No | No | Quick responses, simple queries |
| `--standard` | Yes | No | No | Default, balanced speed and quality |
| `--pro` | Yes | Yes | No | Complex tasks requiring structured planning |
| `--ultra` | Yes | Yes | Yes | Parallel subagent delegation for heavy workloads |

## Features

- **Web Search**: Search the web for current information via Tavily
- **Web Fetch**: Fetch and extract content from web pages via Jina AI
- **Multi-step Reasoning**: Extended thinking for complex problems
- **Planning Mode**: Structured task decomposition with TodoList
- **Subagent Delegation**: Parallel task execution with specialized agents

## Configuration

Requires config.yaml with model credentials. Copy config.example.yaml to config.yaml and configure:

- DEEPSEEK_API_KEY - DeepSeek API key (recommended, cost-effective)
- TAVILY_API_KEY - Tavily API key for web search
- JINA_API_KEY - Jina AI API key for web fetch

Alternative models: OPENAI_API_KEY, ANTHROPIC_API_KEY

## Installation

The skill uses its own embedded deerflow modules. Ensure dependencies are installed:

```bash
pip install langchain langchain-anthropic langchain-openai tavily-python httpx
```

From lgwanai/openforce, MIT.

Signals

Stars
4
Forks
1
Open issues
0
Last commit (this folder)
May 24, 2026 (5 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from lgwanai

Cite this page

deerflow-skill. Agent Mart. https://agentmart.dev/skills/lgwanai-deerflow-skill (last updated 2026-10-24)