Developer telemetry that
actually tells you something.

Real session data from real coding sessions. Built on OpenTelemetry — the same standard used for infrastructure observability — applied to developer productivity.

Telemetry Docs

The pipeline

Standard OTLP from Claude Code to your dashboard. No proprietary agents.

Claude Code

developer sessions

/finish skill

collects spans + metrics

OTEL Collector

standard OTLP/HTTP

ClickHouse

analytics storage

Dashboard

real-time visualization

What gets measured

Every metric comes from real session data — no reconstruction from git timestamps, no surveys.

Session tracking

Automatic time tracking from /morning to /finish. Duration, active time, and idle detection — no manual entry.

Hot files

Files that change in every task are high-risk. Ranked by commit frequency to help prioritize refactoring and documentation.

Tool usage

Which Claude Code tools get used most? Read, Edit, Bash, Search — see adoption patterns and success rates.

Friction detection

Prompt loops and tool failures detected automatically. Surfaced on the Friction Map with severity levels.

Cost metrics

Engineering cost estimates based on session data and team size. Track actual development cost per feature or fix.

Task velocity

How long do similar tasks take? Weekly trends by task type. Helps set realistic expectations during /morning.

Why OpenTelemetry?

No proprietary agents. No vendor lock-in. Standard OTLP protocol means you can route spans and metrics to any backend — Jaeger, Grafana, Datadog, or Tandemu's built-in ClickHouse.

Standard protocol

OTLP/HTTP JSON — the same format used by infrastructure teams worldwide. Nothing proprietary.

Self-hosted pipeline

OTEL Collector, ClickHouse, and the dashboard all run on your infrastructure via Docker Compose.

Extensible

Add custom exporters, route telemetry to your existing observability stack, or build custom dashboards on the raw data.

Developer telemetry thatactually tells you something.