AgentQL is a web extraction and automation toolkit from TinyFish that adds a natural-language query layer on top of live browser sessions. The official repository describes it as a suite that includes Python and JavaScript SDKs, a REST API, a browser debugger, and integrations with tools such as Playwright. Instead of forcing developers to handcraft and maintain a pile of selectors, AgentQL lets them define structured outputs and target elements using higher-level queries tied to the meaning of the page.

That makes it especially relevant for agent workflows where the same job needs to survive layout changes across different sites or repeated runs over time. The upstream docs show common patterns including site login, infinite scroll, product comparison, cookie and popup handling, remote browser use, and structured data extraction. For a team building browsing or scraping agents, those are concrete jobs-to-be-done rather than vague capability claims.

Integration is clear from the source material. The Python SDK installs from PyPI, uses Playwright as the default web driver, requires an AGENTQL_API_KEY, and can be initialized through the CLI or manual setup. The project also has a published npm package and maintained docs, which gives this skill enough source-backed evidence to pass the intake gate. In ASE terms, this belongs in Research & Scraping because it is built for robust web data collection and workflow automation on real pages.