firecrawl-interact

Run firecrawl scrape on the URL first, then use firecrawl interact with a --prompt describing the action, such as clicking a button or filling a form. The scrape ID is saved automatically, so subsequent interact calls target the same session.

Scrape the login page with the --profile flag, then use firecrawl interact to fill in credentials and submit the form. Later scrapes using the same profile reuse the saved cookies and localStorage, keeping you authenticated.

Use interact only when a regular scrape fails because content requires JavaScript interaction, such as clicks, form submissions, pagination, or infinite scroll. It is the last step in the escalation pattern: search, scrape, map, crawl, then interact.

No, interact should never be used for web searches. Use the firecrawl search command instead, which is purpose-built for search queries, while interact is designed for manipulating pages you have already scraped.

Interact requires an active browser session created by a previous firecrawl scrape call. Always scrape the target page first; the scrape ID is saved automatically, or you can pass one explicitly with --scrape-id.