Metrics and alerts for side projects
Log any number.Hear when it spikes, drops or goes quiet.
One API call or one line of code per metric. Spikelog learns what normal looks like for each chart and tells you by email, Slack or Discord when it isn't. No dashboards to build and no rules to write.
Free for 10 charts. No card needed.
checkout_completed dropped to 2 (down 78% vs a typical 9)
Project: Storefront · Open chart · Mute 24h
How it works
- 1
Send a number
From your app, a cron job or a script. The chart appears on the first value.
- 2
Spikelog learns normal
Each chart's usual range and rhythm, from its own recent history.
- 3
You hear about the odd ones
An email with the number and a link, or a message in Slack, Discord or ntfy.
What it catches
The first three are on for every chart by default. You can turn any of them off, set limits, or mute a chart for a while.
Spikes
A value far above what that chart normally does. Error counts, latency, costs.
api_errors spiked to 212 (up 900% vs a typical 21)
Drops
A value far below normal. Signups, checkouts, anything that should keep coming.
signups dropped to 1 (down 88% vs a typical 8)
Gone quiet
A chart that usually gets data every few minutes or hours has stopped. Cron jobs, workers, webhooks.
nightly_backup has gone quiet: no data for 26h (usually arrives within 24h)
Your limits
Optional fixed thresholds for the numbers you already know the rules for.
queue_depth is 1,240, above your limit of 1,000
Works with everything
Client libraries for Node and Python batch in the background and never throw. Anything that can make an HTTP request works too.
curl -X POST https://api.spikelog.com/api/v1/ingest \
-H "X-API-Key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"chart": "Metric Name", "value": 42}'Set up by your coding agent
Copy the AI prompt tab above into Claude Code, Cursor or Copilot. It reads your code, suggests the numbers worth watching (including a heartbeat for each cron job) and adds them.
See Spikelog's own metrics, set up this wayAsk about your numbers
The Spikelog MCP server lets your agent log metrics, list alerts and compare a chart with the day before. Ask “why did signups drop yesterday?” in the same place you fix it.
claude mcp add --transport http spikelog \ https://api.spikelog.com/mcp \ --header "Authorization: Bearer sk_..."
Building in public?
Share a chart or a whole project as a public page, drop a live badge in your README, or embed a sparkline card anywhere that shows images. They update on their own.
Pricing
Start free. Upgrade when a project starts to matter.
Free
$0forever
For side projects and trying things out.
- 10 charts, 1,000 recent points each
- Daily summaries kept forever
- Spike, drop and gone-quiet alerts by email
- Monday digest email
- 1 connected source (GitHub stars, npm or PyPI downloads)
- Public pages, badges and embeds
- MCP server for Claude Code and Cursor
Pro
$6per month
For apps people rely on. Cancel any time from settings.
- 100 charts, 10,000 recent points each
- Alerts to Slack, Discord, ntfy and webhooks
- 20 connected sources
- Badges and embeds without the Spikelog link
- Everything in Free
Public status page
Uptime for the last 90 days, measured every minute.
Your data leaves with you
Export any project as CSV, raw points or daily summaries, whenever you like.
A person answers
Email hello@spikelog.com with anything.
Built for side projects, MVPs and small production apps.
Need logs, traces or high-cardinality queries? Use a full observability tool. Want to know when your numbers do something odd? Use Spikelog.