llms.txt
Compare with other sites' llms.txt files →# Spikelog
> Spikelog stores numeric metrics as charts and alerts when a chart spikes, drops or goes quiet. Charts are created on the first value. Built for side projects and small apps; set it up with one API call per metric.
Base URL: https://api.spikelog.com
Auth for writes: header `X-API-Key: sk_...` (one key per project, from the project settings page in https://spikelog.com/dashboard).
## Send data
- `POST /api/v1/ingest` body `{"chart": "signups", "value": 1}`. Optional `timestamp` (ISO 8601) and `tags` (object, max 4KB). 401 bad key, 422 bad body, 429 chart limit reached.
- `POST /api/v1/ingest/batch` body `{"points": [ ...up to 100 of the above ]}` returns `{"success", "accepted", "rejected": [{"index", "chart", "error"}]}`.
Patterns:
- Measurement: send the actual value (`checkout_ms` = 234).
- Counter: send 1 per event (`signups`); the user can switch the chart to Cumulative mode for a running total.
- Gauge: send the current value on a schedule (`queue_depth`).
- Heartbeat: send 1 every time a cron job or worker runs (`nightly_backup`), so "gone quiet" alerts fire if it stops.
## Client libraries
Node / TypeScript (`npm install spikelog`), zero dependencies, never throws:
```js
import { init, log, increment, gauge, flush } from "spikelog"
init({ apiKey: process.env.SPIKELOG_API_KEY })
increment("signups"); log("checkout_ms", 234); gauge("queue_depth", 12)
await flush() // in serverless functions, before returning
```
Python (`pip install spikelog`), stdlib only:
```python
import spikelog
spikelog.init() # reads SPIKELOG_API_KEY
spikelog.increment("signups"); spikelog.log("checkout_ms", 234); spikelog.gauge("queue_depth", 12)
```
## Alerts (on by default for every chart)
- Spike / drop: value far outside the chart's last 200 points (median + MAD, needs 30 points, and at least 20% from typical).
- Gone quiet: current gap much longer than the chart's usual gap (needs 10 points; minimum 15 minutes).
- Limits: optional fixed above / below values.
Delivered by email; Pro adds Slack, Discord, ntfy and HTTPS webhooks. A digest email goes out on Mondays.
## MCP server
`claude mcp add --transport http spikelog https://api.spikelog.com/mcp --header "Authorization: Bearer sk_..."`
Tools: log_metric, list_charts, get_chart, list_alerts, explain_change (compare the last N hours with the previous N hours), set_alert_limits.
## Public charts
Public charts and projects have SVG badges and sparkline cards:
- `/api/v1/public/badge/c/{chart_slug}.svg?period=7d&label=...`
- `/api/v1/public/spark/c/{chart_slug}.svg?theme=dark&period=30d`
- `/api/v1/public/badge/p/{project_slug}/{chart_name}.svg`
## Limits
Free: 10 charts, 1,000 recent points per chart, email alerts, 1 connected source. Pro ($6/month): 100 charts, 10,000 points, all alert channels, 20 sources. Daily summaries are kept forever on both.
Docs: https://spikelog.com/docs
Example public dashboard (Spikelog's own metrics): https://spikelog.com/p/spikelog
Status: https://spikelog.com/status
Contact: hello@spikelog.com