llama-index-tools-agentql

AgentQL provides web interaction and structured data extraction from any web page using an AgentQL query or a Natural Language prompt. AgentQL can be used across multiple languages and web pages without breaking over time and change.

Warning Only supports async functions and playwright browser APIs, please refer to the following PR for more details: https://github.com/run-llama/llama_index/pull/17808

Installation

pip install llama-index-tools-agentql

You also need to configure the AGENTQL_API_KEY environment variable. You can acquire an API key from our Dev Portal.

Overview

AgentQL provides the following three function tools:

• extract_web_data_with_rest_api: Extracts structured data as JSON from a web page given a URL using either an AgentQL query or a Natural Language description of the data.

• extract_web_data_from_browser: Extracts structured data as JSON from the active web page in a browser using either an AgentQL query or a Natural Language description. This tool must be used with a Playwright browser.

• get_web_element_from_browser: Finds a web element on the active web page in a browser using a Natural Language description and returns its CSS selector for further interaction. This tool must be used with a Playwright browser.

You can learn more about how to use AgentQL tools in this Jupyter notebook.

Extract data using REST API

from llama_index.tools.agentql import AgentQLRestAPIToolSpec

agentql_rest_api_tool = AgentQLRestAPIToolSpec()
await agentql_rest_api_tool.extract_web_data_with_rest_api(
    url="https://www.agentql.com/blog",
    query="{ posts[] { title url author date }}",
)

Work with data and web elements using browser

Setup

In order to use the extract_web_data_from_browser and get_web_element_from_browser, you need to have a Playwright browser instance. If you do not have an active instance, you can initiate one using the create_async_playwright_browser utility method from LlamaIndex's Playwright ToolSpec.

Note AgentQL browser tools are best used along with LlamaIndex's Playwright tools.

from llama_index.tools.playwright.base import PlaywrightToolSpec

async_browser = await PlaywrightToolSpec.create_async_playwright_browser()

You can also use an existing browser instance via Chrome DevTools Protocol (CDP) connection URL:

p = await async_playwright().start()
async_browser = await p.chromium.connect_over_cdp("CDP_CONNECTION_URL")

Extract data from the active browser page

from llama_index.tools.agentql import AgentQLBrowserToolSpec

playwright_tool = PlaywrightToolSpec(async_browser=async_browser)
await playwright_tool.navigate_to("https://www.agentql.com/blog")

agentql_browser_tool = AgentQLBrowserToolSpec(async_browser=async_browser)
await agentql_browser_tool.extract_web_data_from_browser(
    prompt="the blog posts with title and url",
)

Find a web element on the active browser page

next_page_button = await agentql_browser_tool.get_web_element_from_browser(
    prompt="The next page navigation button",
)

await playwright_tool.click(next_page_button)

Agentic Usage

This tool has a more extensive example for agentic usage documented in this Jupyter notebook.

Run tests

In order to run integration tests, you need to configure LLM credentials by setting the OPENAI_API_KEY and AGENTQL_API_KEY environment variables first. Then run the tests with the following command:

make test