langchain-agent-toolbox 0.1.0

LangChain integration for Agent Toolbox API — search, extract, screenshot, weather, finance, email validation, and translation tools for AI agents

🧰 langchain-agent-toolbox

LangChain integration for Agent Toolbox API — 7 tools giving your LangChain agents real-world superpowers.

Tool	Endpoint	Description
AgentToolboxSearchTool	/v1/search	Web search via DuckDuckGo
AgentToolboxExtractTool	/v1/extract	Extract content from any URL
AgentToolboxScreenshotTool	/v1/screenshot	Capture web page screenshots
AgentToolboxWeatherTool	/v1/weather	Current weather & forecasts
AgentToolboxFinanceTool	/v1/finance	Stock quotes & exchange rates
AgentToolboxEmailValidatorTool	/v1/validate-email	Email validation (MX + SMTP + disposable)
AgentToolboxTranslateTool	/v1/translate	Translation with auto-detect & glossary

Installation

pip install langchain-agent-toolbox

Quick Start

Get an API Key

curl -X POST https://api.sendtoclaw.com/v1/auth/register \
  -H "Content-Type: application/json" \
  -d '{"email": "[e-mail]"}'

Set Environment Variable

export AGENT_TOOLBOX_API_KEY="atb_your_key_here"

Use with LangChain

from langchain_agent_toolbox import AgentToolboxSearchTool, AgentToolboxWeatherTool

# Initialize tools
search = AgentToolboxSearchTool()
weather = AgentToolboxWeatherTool()

# Use directly
results = search.invoke({"query": "latest AI news", "count": 3})
print(results)

forecast = weather.invoke({"location": "Tokyo"})
print(forecast)

Use with a LangChain Agent

from langchain_openai import ChatOpenAI
from langchain.agents import AgentExecutor, create_tool_calling_agent
from langchain_core.prompts import ChatPromptTemplate
from langchain_agent_toolbox import (
    AgentToolboxSearchTool,
    AgentToolboxExtractTool,
    AgentToolboxWeatherTool,
    AgentToolboxTranslateTool,
)

# Initialize tools
tools = [
    AgentToolboxSearchTool(),
    AgentToolboxExtractTool(),
    AgentToolboxWeatherTool(),
    AgentToolboxTranslateTool(),
]

# Create agent
llm = ChatOpenAI(model="gpt-4o")
prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant with access to web tools."),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])

agent = create_tool_calling_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

# Run
result = executor.invoke({
    "input": "What's the weather in Paris? Also search for top restaurants there."
})
print(result["output"])

Translation with Glossary

from langchain_agent_toolbox import AgentToolboxTranslateTool

translate = AgentToolboxTranslateTool()

# Simple translation
result = translate.invoke({
    "text": "Hello, how are you?",
    "target": "zh",
})
print(result)  # 你好，你好吗？

# With glossary (preserve technical terms)
result = translate.invoke({
    "text": "The API endpoint returns JSON data.",
    "target": "zh",
    "glossary": {"API": "API", "JSON": "JSON", "endpoint": "端点"},
})
print(result)  # API 端点 返回 JSON 数据。

Email Validation

from langchain_agent_toolbox import AgentToolboxEmailValidatorTool

validator = AgentToolboxEmailValidatorTool()

result = validator.invoke({"email": "[e-mail]"})
print(result)
# {
#   "email": "[e-mail]",
#   "valid_syntax": true,
#   "mx_found": true,
#   "is_disposable": false,
#   "verdict": "deliverable",
#   "score": 0.95
# }

Finance: Stocks & Exchange Rates

from langchain_agent_toolbox import AgentToolboxFinanceTool

finance = AgentToolboxFinanceTool()

# Stock quote
quote = finance.invoke({"symbol": "AAPL"})
print(quote)

# Currency exchange
rate = finance.invoke({
    "from_currency": "USD",
    "to_currency": "EUR",
    "amount": 100,
    "type": "exchange",
})
print(rate)

Configuration

API Key

Set via environment variable (recommended):

export AGENT_TOOLBOX_API_KEY="atb_your_key_here"

Or pass directly:

tool = AgentToolboxSearchTool(api_key="atb_your_key_here")

Custom Base URL

For self-hosted instances:

tool = AgentToolboxSearchTool(
    api_key="your-key",
    base_url="http://localhost:3100",
)

Or via environment variable:

export AGENT_TOOLBOX_BASE_URL="http://localhost:3100"

Async Support

All tools support async out of the box:

import asyncio
from langchain_agent_toolbox import AgentToolboxSearchTool

async def main():
    search = AgentToolboxSearchTool()
    result = await search.ainvoke({"query": "async python", "count": 3})
    print(result)

asyncio.run(main())

API Reference

All tools accept these common parameters:

Parameter	Type	Default	Description
api_key	str	$AGENT_TOOLBOX_API_KEY	API key for authentication
base_url	str	https://api.sendtoclaw.com	API base URL

Links

• Agent Toolbox API — main project
• OpenAPI Spec — full API documentation
• LangChain — LangChain framework

License

MIT