research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Research Skill

Conduct comprehensive research on any topic with automatic source gathering, analysis, and response generation with citations.

对任意主题开展全面研究，自动收集来源、分析并生成带引用的回复。

Prerequisites

前提条件

Tavily API Key Required - Get your key at https://tavily.com

Add to

~/.claude/settings.json

json

{
  "env": {
    "TAVILY_API_KEY": "tvly-your-api-key-here"
  }
}

需要Tavily API Key - 前往https://tavily.com获取你的密钥。

添加到

~/.claude/settings.json

中：

json

{
  "env": {
    "TAVILY_API_KEY": "tvly-your-api-key-here"
  }
}

Quick Start

快速开始

Tip: Research can take 30-120 seconds. Press Ctrl+B to run in the background.

提示：研究过程可能需要30-120秒。按Ctrl+B可在后台运行。

Using the Script

使用脚本

bash

./scripts/research.sh '<json>' [output_file]

Examples:

bash

undefined

bash

./scripts/research.sh '<json>' [output_file]

示例：

bash

undefined

Basic research

./scripts/research.sh '{"input": "quantum computing trends"}'

With pro model for comprehensive analysis

./scripts/research.sh '{"input": "AI agents comparison", "model": "pro"}'

Save to file

./scripts/research.sh '{"input": "market analysis for EVs", "model": "pro"}' ./ev-report.md

With custom citation format

./scripts/research.sh '{"input": "climate change impacts", "model": "mini", "citation_format": "apa"}'

With structured output schema

./scripts/research.sh '{"input": "fintech startups 2025", "model": "pro", "output_schema": {"properties": {"summary": {"type": "string"}, "companies": {"type": "array", "items": {"type": "string"}}}, "required": ["summary"]}}'

undefined

undefined

Basic Research

基础研究

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Latest developments in quantum computing",
    "model": "mini",
    "stream": false,
    "citation_format": "numbered"
  }'

Note: Streaming is disabled for token management. The call waits until research completes and returns clean JSON.

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Latest developments in quantum computing",
    "model": "mini",
    "stream": false,
    "citation_format": "numbered"
  }'

注意：为了管理令牌，流式传输已禁用。调用会等待研究完成后返回整洁的JSON。

With Custom Schema

使用自定义模式

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Electric vehicle market analysis",
    "model": "pro",
    "stream": false,
    "citation_format": "numbered",
    "output_schema": {
      "properties": {
        "market_overview": {
          "type": "string",
          "description": "2-3 sentence overview of the market"
        },
        "key_players": {
          "type": "array",
          "description": "Major companies in this market",
          "items": {
            "type": "object",
            "properties": {
              "name": {"type": "string", "description": "Company name"},
              "market_share": {"type": "string", "description": "Approximate market share"}
            },
            "required": ["name"]
          }
        }
      },
      "required": ["market_overview", "key_players"]
    }
  }'

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Electric vehicle market analysis",
    "model": "pro",
    "stream": false,
    "citation_format": "numbered",
    "output_schema": {
      "properties": {
        "market_overview": {
          "type": "string",
          "description": "2-3 sentence overview of the market"
        },
        "key_players": {
          "type": "array",
          "description": "Major companies in this market",
          "items": {
            "type": "object",
            "properties": {
              "name": {"type": "string", "description": "Company name"},
              "market_share": {"type": "string", "description": "Approximate market share"}
            },
            "required": ["name"]
          }
        }
      },
      "required": ["market_overview", "key_players"]
    }
  }'

API Reference

API参考

Endpoint

端点

POST https://api.tavily.com/research

POST https://api.tavily.com/research

Headers

请求头

Header	Value
`Authorization`	`Bearer <TAVILY_API_KEY>`
`Content-Type`	`application/json`

请求头	值
`Authorization`	`Bearer <TAVILY_API_KEY>`
`Content-Type`	`application/json`

Request Body

请求体

Field	Type	Default	Description
`input`	string	Required	Research topic or question
`model`	string	`"mini"`	Model: `mini` , `pro` , `auto`
`stream`	boolean	`false`	Streaming disabled for token management
`output_schema`	object	null	JSON schema for structured output
`citation_format`	string	`"numbered"`	Citation format: `numbered` , `mla` , `apa` , `chicago`

字段	类型	默认值	描述
`input`	字符串	必填	研究主题或问题
`model`	字符串	`"mini"`	模型： `mini` 、 `pro` 、 `auto`
`stream`	布尔值	`false`	为了管理令牌，流式传输已禁用
`output_schema`	对象	null	用于结构化输出的JSON模式
`citation_format`	字符串	`"numbered"`	引用格式： `numbered` 、 `mla` 、 `apa` 、 `chicago`

Response Format (JSON)

响应格式（JSON）

With

stream: false

, the response is clean JSON:

json

{
  "content": "# Research Results\n\n...",
  "sources": [{"url": "https://...", "title": "Source Title"}],
  "response_time": 45.2
}

当

stream: false

时，响应为整洁的JSON：

json

{
  "content": "# Research Results\n\n...",
  "sources": [{"url": "https://...", "title": "Source Title"}],
  "response_time": 45.2
}

Model Selection

模型选择

Rule of thumb: "what does X do?" -> mini. "X vs Y vs Z" or "best way to..." -> pro.

Model	Use Case	Speed
`mini`	Single topic, targeted research	~30s
`pro`	Comprehensive multi-angle analysis	~60-120s
`auto`	API chooses based on complexity	Varies

经验法则：“X有什么作用？” -> 使用mini模型。“X vs Y vs Z”或“最佳方法是……” -> 使用pro模型。

模型	使用场景	速度
`mini`	单一主题、针对性研究	~30秒
`pro`	全面多角度分析	~60-120秒
`auto`	API根据复杂度自动选择	视情况而定

Schema Usage

模式使用

Schemas make output structured and predictable. Every property MUST include both

type

and

description

json

{
  "properties": {
    "summary": {
      "type": "string",
      "description": "2-3 sentence executive summary"
    },
    "key_points": {
      "type": "array",
      "description": "Main takeaways",
      "items": {"type": "string"}
    }
  },
  "required": ["summary", "key_points"]
}

模式可让输出结构化且可预测。每个属性必须同时包含

type

和

description

。

json

{
  "properties": {
    "summary": {
      "type": "string",
      "description": "2-3 sentence executive summary"
    },
    "key_points": {
      "type": "array",
      "description": "Main takeaways",
      "items": {"type": "string"}
    }
  },
  "required": ["summary", "key_points"]
}

Examples

示例

Market Research

市场研究

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Fintech startup landscape 2025",
    "model": "pro",
    "stream": false,
    "citation_format": "numbered",
    "output_schema": {
      "properties": {
        "market_overview": {"type": "string", "description": "Executive summary of fintech market"},
        "top_startups": {
          "type": "array",
          "description": "Notable fintech startups",
          "items": {
            "type": "object",
            "properties": {
              "name": {"type": "string", "description": "Startup name"},
              "focus": {"type": "string", "description": "Primary business focus"},
              "funding": {"type": "string", "description": "Total funding raised"}
            },
            "required": ["name", "focus"]
          }
        },
        "trends": {"type": "array", "description": "Key market trends", "items": {"type": "string"}}
      },
      "required": ["market_overview", "top_startups"]
    }
  }'

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "Fintech startup landscape 2025",
    "model": "pro",
    "stream": false,
    "citation_format": "numbered",
    "output_schema": {
      "properties": {
        "market_overview": {"type": "string", "description": "Executive summary of fintech market"},
        "top_startups": {
          "type": "array",
          "description": "Notable fintech startups",
          "items": {
            "type": "object",
            "properties": {
              "name": {"type": "string", "description": "Startup name"},
              "focus": {"type": "string", "description": "Primary business focus"},
              "funding": {"type": "string", "description": "Total funding raised"}
            },
            "required": ["name", "focus"]
          }
        },
        "trends": {"type": "array", "description": "Key market trends", "items": {"type": "string"}}
      },
      "required": ["market_overview", "top_startups"]
    }
  }'

Technical Comparison

技术对比

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "LangGraph vs CrewAI for multi-agent systems",
    "model": "pro",
    "stream": false,
    "citation_format": "mla"
  }'

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "LangGraph vs CrewAI for multi-agent systems",
    "model": "pro",
    "stream": false,
    "citation_format": "mla"
  }'

Quick Overview

快速概览

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "What is retrieval augmented generation?",
    "model": "mini",
    "stream": false
  }'

bash

curl --request POST \
  --url https://api.tavily.com/research \
  --header "Authorization: Bearer $TAVILY_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
    "input": "What is retrieval augmented generation?",
    "model": "mini",
    "stream": false
  }'