AgentQL
Natural-language query language for the web: resilient, self-healing web extraction for AI agents.
About
AgentQL, built by TinyFish, is a query language for web pages that replaces brittle CSS selectors and XPath with semantic queries. Developers write GraphQL-style queries in natural language (e.g. add_to_cart_button, product_price) and AgentQL's AI maps them to actual page elements, so queries survive site redesigns because they are based on intent rather than DOM structure.
It ships as Python and JavaScript SDKs with drop-in Playwright integration, a headless browser debugger, and a REST API for agents that are not browser-based. It supports structured data extraction from web pages, PDFs, and images, and integrates with LangChain, LlamaIndex, MCP, Zapier, and Google ADK. Pricing is API-based and usage-metered, with a free Starter tier and Professional at $99/mo.
Limitations: $99/mo Professional is pricier than the median for data-extraction tools, and semantic matching can misfire on highly ambiguous pages where a human-written selector would be exact.
What it can do
- Natural-language element queries that survive site redesigns
- Structured data extraction with GraphQL-style schemas
- Playwright integration plus REST API for headless access
- PDF table parsing and authenticated-site support
Setup
Sign up at agentql.com for API credentials; pip install agentql or npm install agentql, then wrap a Playwright page.
Prerequisites
Python or JavaScript environment; Playwright for browser-mode use.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 80; Trending score 78/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.