fix: add missing params to firecrawl_extract tool (systemPrompt, scrapeOptions, showSources, etc.)

[MaxwellCalkin]

Summary

The firecrawl_extract tool was missing several parameters that both the Firecrawl API backend and JS SDK prepareExtractPayload support:

Parameter	Type	Purpose
systemPrompt	string	Custom system prompt to guide LLM extraction behavior
showSources	boolean	Include source URLs in the response
scrapeOptions	ScrapeOptions	Scrape options applied to each extracted page (reuses existing scrapeParamsSchema)
ignoreInvalidURLs	boolean	Skip invalid/blocked URLs instead of failing the request
ignoreSitemap	boolean	Skip sitemap when discovering URLs
limit	number	Maximum number of URLs to process
timeout	number	Custom timeout in milliseconds (minimum 1000)
agent	{ model: 'FIRE-1' }	Agent options for extraction model selection
webhook	string	Webhook URL for async notifications (gated by SAFE_MODE)
webhookHeaders	Record<string, string>	Custom headers for webhook requests (gated by SAFE_MODE)

Implementation

• Follows the same patterns already used by firecrawl_crawl and firecrawl_search:
  • scrapeOptions reuses scrapeParamsSchema.omit({ url: true }).partial() and goes through transformScrapeParams()
  • webhook/webhookHeaders are gated by SAFE_MODE and processed through buildWebhook()
• Updated the tool description to document all new parameters
• Updated the usage example to show systemPrompt and showSources
• TypeScript compiles cleanly (npx tsc --noEmit passes)

AI Disclosure

This PR was authored by Claude Opus 4.6 (Anthropic), an AI working transparently on behalf of Max Calkin. See https://max.careers for details.