Automate Web Browsing with Natural Language
Browse the web and act on your behalf through Chrome via MultiOn in a LlamaIndex agent.
Why it matters
Empower your AI agent to interact with the internet through your Chrome browser. This tool allows agents to perform actions on your behalf by translating natural language commands into browser operations.
Outcomes
What it gets done
Execute web browsing tasks based on natural language instructions.
Integrate with MultiOn for seamless browser automation.
Enable AI agents to access and act on web content.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/li-tool-tools-multion | bash Overview
MultiOn Tool
A LlamaIndex tool that lets an agent browse and act inside the user's logged-in Chrome browser via the MultiOn extension. Use when an agent needs to read or act on personalized, logged-in browser content that a generic API can't reach.
What it does
The MultiOn Tool connects a LlamaIndex agent to MultiOn, letting the agent connect to the internet through the user's Chrome web browser and act on their behalf. It requires both the MultiOn Chrome extension installed locally and a MultiOn account to function - this is a browser-automation tool, not a headless API-only integration.
MultionToolSpec exposes a single function, browse, which takes natural-language instructions and passes them to the web browser to execute. The source's own worked example shows the agent reading the latest tweets from the user's followers and checking the next event on their Google Calendar - both tasks that require the agent to actually navigate and read real, logged-in browser sessions rather than working from a generic API.
When to use - and when NOT to
Use it when a LlamaIndex agent needs to act inside the user's actual, logged-in browser sessions - reading personalized content like a social feed or calendar, or performing actions a plain API call couldn't reach. Do not use it expecting a headless or server-side browsing tool; it depends on the MultiOn Chrome extension running in the user's own browser, so it only works in a context where that browser and extension are present and set up.
Capabilities
browse takes a natural-language instruction and executes it in the user's Chrome browser via the MultiOn extension, letting the agent read or act on real, personalized browser content.
How to install
pip install llama-index-tools-multion
Requires the MultiOn Chrome extension installed and a MultiOn account.
Who it's for
Developers building LlamaIndex agents that need to act inside a user's real, logged-in browser sessions - reading personalized feeds, calendars, or other content a generic API can't reach.
Source README
MultiOn Tool
pip install llama-index-tools-multion
This tool connects to MultiOn to enable your agent to easily
connect to the internet through your Chrome Web browser and act on your behalf
You will need to have the MultiOn chrome extension installed and a MultiOn account
to use this integration
Usage
This tool has more a extensive example usage documented in a Jupyter notebook here.
Here's an example usage of the MultionToolSpec.
from llama_index.tools.multion import MultionToolSpec
from llama_index.core.agent.workflow import FunctionAgent
from llama_index.llms.openai import OpenAI
multion_tool = MultionToolSpec(api_key="your-multion-key")
agent = FunctionAgent(
tools=multion_tool.to_tool_list(),
llm=OpenAI(model="gpt-4.1"),
)
print(await agent.run("Can you read the latest tweets from my followers"))
print(await agent.run("What's the next thing on my google calendar?"))
browse: The core function that takes natural language instructions to pass to the web browser to execute
This loader is designed to be used as a way to load data as a Tool in a Agent.
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.