Appearance
AgentQL
Category: Framework
Pricing: Free trial (300 API calls, 1 hr remote browser); Starter: $0/mo with 50 free calls then $0.02/call; Professional: $99/mo (10,000 calls included); Enterprise: custom
Official site: https://www.agentql.com/
Docs: https://docs.agentql.com/
GitHub: https://github.com/tinyfish-io/agentql
Playground: https://playground.agentql.com/
What it is
AgentQL is a query language and SDK that connects LLMs and AI agents to the web. Instead of fragile CSS selectors or XPath, you describe the element you want in plain language and AgentQL finds it using AI-powered semantic matching. It wraps a Playwright page object and adds two methods: queryElements (for interaction targets like buttons and inputs) and queryData (for structured data extraction).
It works on any page, public or private, including pages behind authentication, because it operates inside a real browser session.
Core use cases:
- Web scraping without brittle selectors (e-commerce, job boards, social media)
- Browser automation where the DOM changes frequently
- Data extraction from authenticated or dynamic pages
- AI agent pipelines that need reliable web access
- PDF parsing and table extraction via the REST API
Key capabilities
| Feature | Notes |
|---|---|
| Semantic element selection | Natural language queries replace CSS/XPath selectors |
| Self-healing selectors | Adapts automatically when the DOM changes |
| Python and JavaScript SDKs | Both wrap Playwright; same query language |
| REST API | Browserless structured data extraction without running a browser |
| Browser extension debugger | Chrome extension for testing queries live in the browser |
| Authenticated pages | Works on private pages inside an existing browser session |
| Remote browser option | Professional plan includes managed remote browser hours |
Safety and ToS
AgentQL is MIT-licensed open-source software (the SDK). An API key from https://dev.agentql.com/ is required for all requests. Usage is metered against your plan's call quota.
Standard web scraping ethics apply: respect robots.txt, do not hammer targets with high-frequency requests, and comply with the target site's own terms of service. AgentQL's Terms of Service are at https://www.agentql.com/legal/terms-of-service.