API-first analytics for autonomous AI systems. Your agent tracks events and queries its own performance โ no browser, no SDK, just HTTP.
We welcome all bots, agents, and autonomous systems.
๐ค ๐ฆพ ๐ง ๐ฆ โ๏ธ ๐งTraditional analytics are dashboards for humans. Your agent can't log into Mixpanel. With Agent Analytics, it queries its own stats via API โ and acts on them.
Sub-5ms response. Track events without slowing down your agent's workflow. Fire-and-forget โ we persist in the background.
Runs on Cloudflare's global network. Low latency from anywhere your agent runs โ cloud, local, container, Raspberry Pi.
Events, properties, timestamps, users. That's it. No session stitching, no funnels, no complexity you don't need.
Self-hosted on Cloudflare free tier: 100K requests/day, 5GB storage. Most agents never outgrow this.
If it can make an HTTP POST, it can use Agent Analytics. Python, Node, Bash, Go, Rust โ one curl command.
Traditional analytics tools assume a browser. They load JavaScript SDKs, set cookies, track page views. None of that makes sense for an AI agent running in a terminal.
Agent Analytics is pure HTTP. Your agent can track every action it takes โ tasks completed, errors hit, tools used, time spent โ then query that data to improve itself.
# Track an action
curl -X POST https://api.agentanalytics.sh/track \
-H "Content-Type: application/json" \
-d '{
"project": "my-agent",
"event": "task_completed",
"properties": {
"task": "send_report",
"duration_ms": 1250
}
}'
# Query your stats (agents can do this!)
curl "https://api.agentanalytics.sh/stats\
?project=my-agent&days=7" \
-H "X-API-Key: $KEY"
# Response:
# { "totals": { "total_events": 2847,
# "unique_users": 1 },
# "events": [
# { "event": "task_completed",
# "count": 1923 } ] }
We only store what you explicitly send. No fingerprinting, no PII collection, no sneaky scripts.
Read access requires your API key. Write access is open by design โ agents need frictionless tracking.
Open source codebase. Both self-hosted and cloud run the same code. Read every line before deploying.