llamaindex-agent-toolbox 0.1.0

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

🧰 llamaindex-agent-toolbox

LlamaIndex integration for Agent Toolbox API — 7 tools giving your LlamaIndex 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 llamaindex-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 Tools Directly

from llamaindex_agent_toolbox import AgentToolboxSearchTool, AgentToolboxWeatherTool

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

# Call directly
result = search.call(query="latest AI news", count=3)
print(result.content)

result = weather.call(location="Tokyo")
print(result.content)

Use with a LlamaIndex Agent

from llama_index.core.agent import ReActAgent
from llama_index.llms.openai import OpenAI
from llamaindex_agent_toolbox import (
    AgentToolboxSearchTool,
    AgentToolboxExtractTool,
    AgentToolboxWeatherTool,
    AgentToolboxTranslateTool,
)

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

# Create agent
llm = OpenAI(model="gpt-4o")
agent = ReActAgent.from_tools(tools, llm=llm, verbose=True)

# Run
response = agent.chat("What's the weather in Paris? Translate the summary to Japanese.")
print(response)

Translation with Glossary

from llamaindex_agent_toolbox import AgentToolboxTranslateTool

translate = AgentToolboxTranslateTool()

# Simple translation
result = translate.call(text="Hello, how are you?", target="zh")
print(result.content)  # 你好，你好吗？

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

Email Validation

from llamaindex_agent_toolbox import AgentToolboxEmailValidatorTool

validator = AgentToolboxEmailValidatorTool()
result = validator.call(email="[e-mail]")
print(result.content)
# {"email": "[e-mail]", "verdict": "deliverable", "score": 0.95, ...}

Async Support

All tools support async via acall:

import asyncio
from llamaindex_agent_toolbox import AgentToolboxSearchTool

async def main():
    search = AgentToolboxSearchTool()
    result = await search.acall(query="async python", count=3)
    print(result.content)

asyncio.run(main())

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"

Links

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

License

MIT