TinyFish tools

Start an asynchronous browser agent on a URL and have it work toward a natural-language goal.

Search research papers by topic, publication year, domain, language, and date.

Search TinyFish's news index with optional recency, date, domain, language, and location filters.

Search the web by keyword or meaning, with domain, language, location, and date filters.

Fetches web pages, renders JavaScript-heavy pages when needed, and returns clean extracted content in your preferred format (Markdown, cleaned semantic HTML, or a structured JSON tree), with optional links, metadata, and selectors.

Render pages in a real browser and return clean, model-ready content for 1-10 URLs in a single call, fetched in parallel. 'format' selects markdown (default), semantic html, or a structured json document tree. Scope extraction with CSS selectors: 'include_selectors' keeps only matching regions and returns 'candidate_selectors' retry hints on a total miss rather than silently falling back to the whole page, while 'exclude_selectors' prunes boilerplate first. 'ttl' controls cache freshness — omit it for any cached copy, 0 to force a live fetch, or a maximum age in seconds. To poll a page for changes cheaply, fetch it once with 'include_etag_and_last_modified' AND 'ttl' of 0 to capture its 'etag' / 'last_modified', then replay those as 'if_none_match' / 'if_modified_since' on later calls: an unchanged page comes back as 'not_modified: true' with no body at all. Optional 'links' and 'image_links' return absolute URLs. HTML, PDF, JSON, and plain text are supported; images are not. A per-URL

Search the live web, news, or research papers and get ranked results.