Ai

Firecrawl v2.9.0 Adds Interactive Browser Automation and Query-Based Scraping

Firecrawl v2.9.0 Adds Interactive Browser Automation and Query-Based Scraping

Firecrawl v2.9.0 expands the platform from straightforward web scraping into more interactive browser-driven workflows. The release introduces a new /interact endpoint for taking actions on pages after scraping, alongside new output formats, improved document parsing, and broader SDK coverage. For teams building AI agents, research tools, or automation pipelines, this version makes it easier to work with dynamic sites and extract cleaner, more targeted results.

What Changed

The headline feature in v2.9.0 is the new /interact endpoint. After scraping a page, users can now continue working within a persistent browser session to click buttons, fill forms, navigate deeper into sites, or retrieve dynamic content. Firecrawl supports both natural-language prompts and lower-level scripted control through Playwright code or Bash with agent-browser. The addition of persistent sessions, live view streaming, and reusable browser profiles also gives developers more control over authenticated or stateful workflows.

The /scrape endpoint also gains a new query format, allowing users to submit a natural-language prompt and receive a direct answer in data.answer. This reduces the need for post-processing in use cases where the goal is not just page capture, but a concise extracted answer.

Another addition is the new audio format, which returns audio output as part of the document response. This broadens the kinds of downstream experiences developers can build, especially for accessibility or voice-first workflows.

For cleaner extraction, Firecrawl now includes the onlyCleanContent parameter on /scrape. This option strips out navigation elements, advertisements, cookie banners, and other non-semantic page clutter from markdown output, helping teams work with more focused content.

Document ingestion also improves in v2.9.0. PDF parsing now supports fast, auto, and ocr modes, along with a maxPages option to control how deeply files are processed. Firecrawl also adds support for legacy .doc files, extending compatibility for enterprise document workflows.

On the developer tooling side, the release introduces official Java and Elixir SDKs with full v2 API support, building on existing language coverage across JavaScript, Python, Rust, and Java integrations mentioned in the release notes. Firecrawl also adds a dedicated Wikimedia scraping engine for Wikipedia and related properties.

Why It Matters

This update is significant because it pushes Firecrawl beyond static extraction into interactive web task execution. The new browser session model can support a wider class of AI and automation use cases, particularly where websites require clicks, form submissions, or stateful browsing before valuable content becomes available.

The query and onlyCleanContent additions also reflect a broader trend toward more answer-oriented and model-friendly data pipelines. Instead of returning large volumes of raw page data, Firecrawl is giving developers tools to request cleaner and more purposeful outputs directly from the API.

For enterprise teams, the expanded SDK support and stronger document parsing controls may reduce integration friction. Java and Elixir support opens the platform to more backend environments, while the new PDF modes and legacy document support improve handling of mixed-format business content.

Overall, Firecrawl v2.9.0 is a meaningful product update for developers building AI agents, browser automation systems, and knowledge extraction pipelines. Its biggest impact will likely come from the new interactive scraping workflow, which makes it easier to bridge the gap between page retrieval and real browser-based task execution.

Official Source: https://github.com/firecrawl/firecrawl/releases/tag/v2.9.0