Mastering Firecrawl's interact endpoint: a complete guide
Blog post from Firecrawl
Firecrawl is an API designed to provide AI agents with web context through its /scrape and /interact endpoints, enabling interactions such as search, scraping, and live-browser engagements. The /scrape endpoint fetches a webpage and returns its content in formats like markdown or structured data, while /interact keeps a live browser session open, allowing for complex interactions such as logins or multi-step flows. Users can drive these sessions using plain-English prompts or code in Node, Python, or Bash, with the ability to mix modes within the same session. Named profiles allow the retention of session states, such as cookies and login information, across different scrapes. Billing is calculated per session-minute, with different rates for code-only and prompt-based interactions. This infrastructure is particularly beneficial for accessing web data that requires interactions beyond a simple HTTP request, such as content behind logins or pages requiring multiple steps to reveal information.