Claude Code Integration

Claude Code already exports rich OpenTelemetry data — interactions, LLM requests, tool executions, token usage, and billing signals. Traccia ingests that data natively on the hosted platform. Point Claude Code at Traccia, run your session, and open the dashboard to see exactly what happened: which tools ran, how much context grew, what it cost, and how multi-turn conversations roll up into sessions.

1Get a Traccia API key

Sign in to app.traccia.ai, open Settings → API Keys, and create an ingestion key. Copy the key — you will paste it into Claude Code's settings in the next step.

2Configure Claude Code

Add one of the following to ~/.claude/settings.json. Replace YOUR_TRACCIA_API_KEY with your key. Set env to match your environment (e.g. production, staging).

HTTP (protobuf)

{
  "env": {
    "CLAUDE_CODE_ENABLE_TELEMETRY": "1",
    "OTEL_EXPORTER_OTLP_PROTOCOL": "http/protobuf",
    "OTEL_EXPORTER_OTLP_TRACES_ENDPOINT": "https://api.traccia.ai/v2/traces",
    "OTEL_EXPORTER_OTLP_METRICS_ENDPOINT": "https://api.traccia.ai/v2/metrics",
    "OTEL_EXPORTER_OTLP_LOGS_ENDPOINT": "https://api.traccia.ai/v2/logs",
    "OTEL_EXPORTER_OTLP_HEADERS": "Authorization=Bearer YOUR_TRACCIA_API_KEY",
    "OTEL_SERVICE_NAME": "claude-code",
    "OTEL_RESOURCE_ATTRIBUTES": "agent.id=claude-code,agent.name=claude-code,env=production"
  }
}

gRPC

Claude Code's default OTLP transport. Uses a single endpoint for traces, metrics, and logs. For the general gRPC reference (all native OTel clients), see Platform OTLP Ingestion.

{
  "env": {
    "CLAUDE_CODE_ENABLE_TELEMETRY": "1",
    "OTEL_EXPORTER_OTLP_PROTOCOL": "grpc",
    "OTEL_EXPORTER_OTLP_ENDPOINT": "https://api.traccia.ai:4317",
    "OTEL_EXPORTER_OTLP_HEADERS": "Authorization=Bearer YOUR_TRACCIA_API_KEY",
    "OTEL_SERVICE_NAME": "claude-code",
    "OTEL_RESOURCE_ATTRIBUTES": "agent.id=claude-code,agent.name=claude-code,env=production"
  }
}

3Run Claude Code and open Traccia

Start a session as you normally would:

claude

Ask Claude Code to do real work — read files, run commands, iterate on a task. Telemetry exports every few seconds and flushes when you exit cleanly.

In Traccia, go to Traces, filter by agent claude-code, and click any trace to open the full timeline. Your first trace usually appears within seconds of completing a prompt.

Sessions

Claude Code assigns every conversation a stable session.id. Each time you send a new prompt in the same terminal session, Claude Code creates a new interaction (trace) that shares that session ID.

In Traccia, turn on Group by session on the Traces page to collapse a conversation into one row:

• Session row — total tokens, duration, billed cost, and CC reference cost rolled up across all interactions
• Interaction rows — individual prompts expanded underneath
• Open — session detail page with every interaction in chronological order
• Badges — interaction count and tool retry count at a glance

Understanding cost

Traccia labels costs based on how Claude Code is routed:

Agent identity

Keep agent.id=claude-code in OTEL_RESOURCE_ATTRIBUTES so every session rolls up under one agent in the dashboard.

For teams running multiple CC profiles (e.g. different repos or environments), use distinct agent.id values like claude-code-backend and claude-code-frontend, and set env to match.

Roll out to your team

For organization-wide rollout, use Claude Code managed settings to inject the same env block for every developer. Pair each team with a Traccia workspace and ingestion key scoped to that workspace so traces stay isolated.

Troubleshooting