Browser API for AI agents
4 live endpoints on UnifAPI · OpenAPI-published operation billing · one public-data layer for Skills and custom agent workflows.
Render any URL in a headless browser — JavaScript executed — and get it back as clean Markdown, fully rendered HTML, a screenshot, or the page's link graph. The web-reading primitives an SEO or marketing agent needs to read pages that plain fetches can't.
Use Browser records as evidence, not glue code.
UnifAPI keeps the public Browser surface discoverable through MCP and callable over HTTP. Agents can inspect the schema, fetch only the records needed for the task, and return the result with request, pagination, and billing metadata intact.
Fetch public Browser records when an agent needs source evidence instead of generic web text.
Use the same operation contracts through MCP for exploratory Skills and through HTTP for production integrations.
Combine this platform with the rest of UnifAPI's catalog without changing response envelopes or billing logic.
Agents & Skills that use the Browser API
You rarely call an API by hand. These prebuilt agents and skills already read Browser records inside Claude or Codex — start from one of them, or call the endpoints below yourself.
4 ready-to-call endpoints
Render a page to HTML
Render a URL in a headless browser and return the fully rendered HTML after JavaScript execution — use it to read meta tags, structured data, and content that only appears after client-side rendering.
Extract links from a page
Render a URL in a headless browser and return every link on the page as absolute URLs, including links injected by JavaScript. Optionally restrict to visible links or same-domain links for internal-link and broken-link audits.
Render a page to Markdown
Render a URL in a headless browser (running JavaScript) and return the page as clean Markdown — ideal for content analysis, summarization, and feeding pages to language models.
Capture a page screenshot
Render a URL in a headless browser and capture a screenshot, returned as a base64-encoded image. Control the viewport, capture the full scrollable page, target a CSS selector, and choose png, jpeg, or webp output.