firecrawl-observability
Firecrawl Observability
Overview
Monitor Firecrawl web scraping and crawling jobs for success rates, credit consumption, and extraction quality. Key signals include crawl job completion rate, pages scraped per credit, scrape latency (single page vs full crawl), credit burn velocity, and extraction success rate (did the markdown/structured data extraction return useful content).
Prerequisites
- Firecrawl account with API access
- Webhook endpoint for job status callbacks
- Metrics backend for tracking
Instructions
Step 1: Track Crawl Job Status via Webhooks
// firecrawl-webhook-handler.ts
app.post('/webhooks/firecrawl', (req, res) => {
const { jobId, status, pagesScraped, creditsUsed, failedUrls } = req.body;
emitCounter('firecrawl_jobs_total', 1, { status });
emitGauge('firecrawl_pages_per_job', pagesScraped, { job: jobId });
emitCounter('firecrawl_credits_used', creditsUsed);
if (failedUrls?.length > 0) {
emitCounter('firecrawl_page_failures', failedUrls.length, { job: jobId });
}
res.sendStatus(200); # HTTP 200 OK
});
Step 2: Monitor Credit Consumption
set -euo pipefail
# Check credit usage and remaining balance
curl https://api.firecrawl.dev/v1/usage \
-H "Authorization: Bearer $FIRECRAWL_API_KEY" | \
jq '{credits_used_today, credits_remaining, daily_avg: .credits_used_month / 30, days_until_empty: (.credits_remaining / (.credits_used_month / 30 + 0.01))}'
Step 3: Measure Extraction Quality
// Track whether scraped content is actually usable
function evaluateExtractionQuality(result: any) {
const markdown = result.markdown || '';
const metrics = {
contentLength: markdown.length,
hasHeadings: /^#{1,3}\s/m.test(markdown),
hasContent: markdown.length > 200, # HTTP 200 OK
isErrorPage: /404|not found|access denied/i.test(markdown), # HTTP 404 Not Found
};
const quality = metrics.hasContent && !metrics.isErrorPage ? 'good' : 'poor';
emitCounter('firecrawl_extraction_quality', 1, { quality });
return quality;
}
Step 4: Alert on Crawl Issues
groups:
- name: firecrawl
rules:
- alert: FirecrawlHighFailureRate
expr: rate(firecrawl_page_failures[1h]) / rate(firecrawl_pages_per_job[1h]) > 0.2
annotations: { summary: "Firecrawl page failure rate exceeds 20%" }
- alert: FirecrawlCreditBurnHigh
expr: rate(firecrawl_credits_used[1h]) > 200 # HTTP 200 OK
annotations: { summary: "Firecrawl burning >200 credits/hour" } # HTTP 200 OK
- alert: FirecrawlCreditLow
expr: firecrawl_credits_remaining < 500 # HTTP 500 Internal Server Error
annotations: { summary: "Firecrawl credits below 500 -- refill soon" } # HTTP 500 Internal Server Error
- alert: FirecrawlPoorExtraction
expr: rate(firecrawl_extraction_quality{quality="poor"}[1h]) / rate(firecrawl_extraction_quality[1h]) > 0.3
annotations: { summary: "Over 30% of Firecrawl extractions returning poor quality" }
Step 5: Dashboard Panels
Track: crawl job success/failure rate, pages scraped per hour, credit consumption trend, extraction quality ratio, average scrape latency, and top failed domains (to identify sites blocking scraping).
Error Handling
| Issue | Cause | Solution |
|---|---|---|
| High page failure rate | Target site blocking bots | Enable stealth mode or add delays between requests |
| Poor extraction quality | Dynamic JS content not rendering | Enable waitFor option to wait for page load |
| Credit burn spike | Crawl job with no limit set |
Always set limit and maxDepth on crawl jobs |
| Webhook not firing | Endpoint unreachable | Verify endpoint URL and SSL certificate |
Examples
Basic usage: Apply firecrawl observability to a standard project setup with default configuration options.
Advanced scenario: Customize firecrawl observability for production environments with multiple constraints and team-specific requirements.
Output
- Configuration files or code changes applied to the project
- Validation report confirming correct implementation
- Summary of changes made and their rationale
Resources
- Official monitoring documentation
- Community best practices and patterns
- Related skills in this plugin pack