Your app went viral.
Your bill went with it.
You built a Lovable app. It got posted on Twitter. Traffic spiked 100x. Your Cloudflare bill went from $5 to $500 overnight. Cloudflare's dashboard didn't tell you for 15 minutes — by then, the damage was done. FlareLog alerts you in real-time, before the invoice lands, before you panic.
How vibe coders get surprise bills
KV write loop
Your AI generated a loop that writes to KV on every request instead of caching. 100k requests = 100k KV writes = $5 in one day. You didn't know KV writes cost money.
D1 query per request
Your app runs a D1 query on every page load instead of caching the result. Viral traffic = millions of D1 row reads. Cloudflare charges per-row-read. Surprise.
Durable Object fan-out
Your WebSocket app creates a new Durable Object per connection. 10k connections = 10k DOs = 10k the cost. The AI didn't mention hibernation.
R2 upload storm
Your file upload feature stores every file in R2. Class A operations (writes) cost $4.50/million. A bot hitting your upload endpoint 1000 times = $4.50 gone in minutes.
FlareLog alerts you before the invoice
FlareLog's Tail Worker estimates the cost of every request in real-time. When your spend crosses a threshold you set, we ping you instantly:
- - Slack notification with the spike source (which endpoint, which resource)
- - Discord webhook for team alerts
- - Custom HTTP webhook for PagerDuty / email / whatever
- - Daily spend summary so you always know where you stand
- - Rate-of-change alerts ("KV writes spiked 5x in 10 min")
The 15-minute lag problem
Cloudflare's own dashboard updates cost metrics with a 15-30 minute delay. That means:
2:00 AM — FlareLog detects the spike. Alerts Slack instantly.
2:01 AM — You wake up. You fix the caching bug. You deploy.
2:15 AM — Cloudflare's dashboard finally shows the spike.
2:16 AM — You already fixed it 15 minutes ago.
Set up cost alerts in 3 steps
# 1. Add your destination in the dashboard
# Projects → your project → Settings → Alert Webhooks
# Paste a Slack, Discord, or any HTTPS endpoint URL.
# You'll get a signing secret to verify deliveries with.
# 2. Enable cost tracking in your Tail Worker config
# In wrangler.toml:
# [vars]
# TRACK_COST = "true"
#
# 3. Set your threshold and spike multiplier
# Projects → your project → Settings → Cost Burn AlertsFAQ
How does FlareLog know my Cloudflare costs?
FlareLog's Tail Worker estimates costs based on the execution outcome of each request — Worker invocations, KV reads/writes, R2 operations, D1 queries. We use Cloudflare's public pricing to calculate real-time cost estimates. The estimates are approximate but close enough to catch runaway spending before it hits your invoice.
What's the '15-minute lag'?
Cloudflare's own dashboard updates cost metrics with a 15-30 minute delay. If your app spikes at 2:00 AM, you won't see it in Cloudflare until 2:15-2:30 AM. FlareLog alerts you within seconds because the Tail Worker processes each request in real-time.
Can I set a custom alert threshold?
Yes. Set a per-request cost threshold (e.g. $0.005) and FlareLog will ping your Slack/Discord/webhook when a single request crosses it. You can also set a spike multiplier (e.g. 'alert me if a request costs 5x the recent average').
Does this work if I'm on the Cloudflare free tier?
Yes. FlareLog tracks your usage against the free tier limits (100k Worker requests/day, 100k KV reads/day, etc.) and alerts you before you hit the cap. If you're on the paid plan, it tracks against your included + billable usage.