Claude Code Monitoring & Observability with OpenTelemetry

SigNoz has released a guide for monitoring Anthropic's Claude Code AI assistant using OpenTelemetry, enabling engineering teams to export logs and metrics on token usage, costs, session activity, and tool decisions to SigNoz dashboards. The integration requires a SigNoz account and Claude Code installation, with setup options for both VSCode and terminal environments using environment variables or bash scripts. Administrators can centrally configure OpenTelemetry settings across their organization through managed settings files for unified visibility into AI usage patterns.

Overview Claude Code monitoring with OpenTelemetry gives you full visibility into how Claude is used across your engineering team. This guide walks you through exporting Claude Code logs and metrics to SigNoz, so you can track the signals that matter most: Token usage & costs — break down spend by user, model, or subagent Session & request activity — track adoption, active time, and request volumes Performance — monitor API latency, cache hit rates, and tool execution times Quota & limits — stay ahead of rate limits before they impact developers Tool decisions — see accept/reject rates across Edit, Write, and other tools Errors & retries — detect API failures and retry exhaustion before they snowball Once set up, all of this flows into SigNoz dashboards where you can correlate logs and metrics, set alerts, and analyze trends over time. Prerequisites - SigNoz setup choose one : SigNoz Cloud account https://signoz.io/teams/ with an active ingestion key- Self-hosted SigNoz instance - Internet access to send telemetry data to SigNoz Cloud Claude Code https://www.anthropic.com/claude-code installed and running on your system Monitoring Claude Code Check out detailed instructions on how to set up OpenTelemetry instrumentation for your Claude Code usage over here https://docs.anthropic.com/en/docs/claude-code/monitoring-usage . Option 1 VSCode Step 1: Launch VSCode with telemetry enabled CLAUDE CODE ENABLE TELEMETRY=1 \ OTEL METRICS EXPORTER=otlp \ OTEL LOGS EXPORTER=otlp \ OTEL EXPORTER OTLP PROTOCOL=grpc \ OTEL EXPORTER OTLP ENDPOINT="https://ingest.<region .signoz.cloud:443" \ OTEL EXPORTER OTLP HEADERS="signoz-ingestion-key=<your-ingestion-key " \ OTEL METRIC EXPORT INTERVAL=10000 \ OTEL LOGS EXPORT INTERVAL=5000 \ code . : Your SigNoz Cloud <region region : Your SigNoz <your-ingestion-key ingestion key Using self-hosted SigNoz? Most steps are identical. To adapt this guide, update the endpoint and remove the ingestion key header as shown in Cloud → Self-Hosted https://signoz.io/docs/ingestion/cloud-vs-self-hosted/ cloud-to-self-hosted . This will open VSCode with the required environment variables already configured. From here, any Claude Code activity will automatically generate telemetry and export logs to your SigNoz Cloud instance. For convenience, you can also clone our bash script https://github.com/SigNoz/Claude-Code-OpenTelemetry/blob/main/claude code otel vscode.sh , update it with your SigNoz endpoint and ingestion key, and run it directly. Option 2 Terminal Step 1: Launch Claude Code with telemetry enabled CLAUDE CODE ENABLE TELEMETRY=1 \ OTEL METRICS EXPORTER=otlp \ OTEL LOGS EXPORTER=otlp \ OTEL EXPORTER OTLP PROTOCOL=grpc \ OTEL EXPORTER OTLP ENDPOINT="https://ingest.<region .signoz.cloud:443" \ OTEL EXPORTER OTLP HEADERS="signoz-ingestion-key=<your-ingestion-key " \ OTEL METRIC EXPORT INTERVAL=10000 \ OTEL LOGS EXPORT INTERVAL=5000 \ claude - Set the to match your SigNoz Cloud <region region - Replace with your SigNoz <your-ingestion-key ingestion key Using self-hosted SigNoz? Most steps are identical. To adapt this guide, update the endpoint and remove the ingestion key header as shown in Cloud → Self-Hosted https://signoz.io/docs/ingestion/cloud-vs-self-hosted/ cloud-to-self-hosted . This will launch Claude Code with telemetry enabled. Any Claude Code activity in the terminal session will automatically generate and export logs and metrics to your SigNoz Cloud instance. For convenience, you can also clone our bash script https://github.com/SigNoz/Claude-Code-OpenTelemetry/blob/main/claude code otel terminal.sh , update it with your SigNoz endpoint and ingestion key, and run it directly. Administrator Configuration Administrators can configure OpenTelemetry settings for all users through the managed settings file. This allows for centralized control of telemetry settings across an organization. See the settings precedence https://docs.anthropic.com/en/docs/claude-code/settings settings-precedence for more information about how settings are applied. The managed settings file is located at: - macOS: /Library/Application Support/ClaudeCode/managed-settings.json - Linux and WSL: /etc/claude-code/managed-settings.json - Windows: C:\ProgramData\ClaudeCode\managed-settings.json Example managed settings configuration: { "env": { "CLAUDE CODE ENABLE TELEMETRY": "1", "OTEL METRICS EXPORTER": "otlp", "OTEL LOGS EXPORTER": "otlp", "OTEL EXPORTER OTLP PROTOCOL": "grpc", "OTEL EXPORTER OTLP ENDPOINT": "http://collector.company.com:4317", "OTEL EXPORTER OTLP HEADERS": "Authorization=Bearer company-token" } } Managed settings can be distributed via MDM Mobile Device Management or other device management solutions. Environment variables defined in the managed settings file have high precedence and cannot be overridden by users. Example Configurations Console debugging 1-second intervals export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=console export OTEL METRIC EXPORT INTERVAL=1000 OTLP/gRPC export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=otlp export OTEL EXPORTER OTLP PROTOCOL=grpc export OTEL EXPORTER OTLP ENDPOINT=http://localhost:4317 Prometheus export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=prometheus Multiple exporters export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=console,otlp export OTEL EXPORTER OTLP PROTOCOL=http/json Different endpoints/backends for metrics and logs export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=otlp export OTEL LOGS EXPORTER=otlp export OTEL EXPORTER OTLP METRICS PROTOCOL=http/protobuf export OTEL EXPORTER OTLP METRICS ENDPOINT=http://metrics.company.com:4318 export OTEL EXPORTER OTLP LOGS PROTOCOL=grpc export OTEL EXPORTER OTLP LOGS ENDPOINT=http://logs.company.com:4317 Metrics only no events/logs export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=otlp export OTEL EXPORTER OTLP PROTOCOL=grpc export OTEL EXPORTER OTLP ENDPOINT=http://localhost:4317 Events/logs only no metrics export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL LOGS EXPORTER=otlp export OTEL EXPORTER OTLP PROTOCOL=grpc export OTEL EXPORTER OTLP ENDPOINT=http://localhost:4317 Your Claude Code activity should now automatically emit logs and metrics. Finally, you should be able to view logs in Signoz Cloud under the logs tab: When you click on any of these logs in SigNoz, you'll see a detailed view of the log, including attributes: You should be able to see Claude Code related metrics in Signoz Cloud under the metrics tab: When you click on any of these metrics in SigNoz, you'll see a detailed view of the metric, including attributes: Dashboard You can also check out our custom Claude Code dashboard here https://signoz.io/docs/dashboards/dashboard-templates/claude-code-dashboard/ which provides specialized visualizations for monitoring your Claude Code usage. The dashboard includes pre-built charts specifically tailored for LLM usage, along with import instructions to get started quickly. Telemetry Data Claude Code emits the following metrics and events via OpenTelemetry. For the full upstream specification, see the Claude Code monitoring docs https://docs.anthropic.com/en/docs/claude-code/monitoring-usage . Standard Attributes Every metric and event includes these attributes: | Attribute | Description | |---|---| session.id | Unique session identifier | user.id | Anonymous device/installation identifier | user.email | User email address when authenticated via OAuth | user.account uuid | Account UUID when authenticated | user.account id | Account ID matching Anthropic admin APIs when authenticated | organization.id | Organization UUID when authenticated | terminal.type | Terminal type, e.g. vscode , iTerm.app , cursor , tmux | app.version | Claude Code version opt-in via OTEL METRICS INCLUDE VERSION=true | Metrics | Metric | Description | Unit | Key Attributes | |---|---|---|---| claude code.session.count | Sessions started | count | start type fresh , resume , continue | claude code.token.usage | Tokens used per API request | tokens | type input , output , cacheRead , cacheCreation , model , query source , speed | claude code.cost.usage | Estimated cost per API request | USD | model , query source , speed , agent.name , skill.name , effort | claude code.lines of code.count | Lines of code added or removed | count | type added , removed | claude code.commit.count | Git commits created | count | — | claude code.pull request.count | Pull requests or merge requests created | count | — | claude code.code edit tool.decision | Edit/Write/NotebookEdit permission decisions | count | tool name , decision accept / reject , source , language | claude code.active time.total | Active usage time | seconds | type user for keyboard input, cli for tool/AI response time | Events Claude Code emits these events via the OpenTelemetry logs/events protocol when OTEL LOGS EXPORTER is configured : | Event | Emitted When | Key Attributes | |---|---|---| claude code.user prompt | User submits a prompt | prompt length , command name , command source , prompt redacted by default — enable with OTEL LOG USER PROMPTS=1 | claude code.api request | API request to Claude completes | model , cost usd , duration ms , input tokens , output tokens , cache read tokens , cache creation tokens , request id , speed , query source | claude code.api error | API request fails after all retries | model , error , status code , duration ms , attempt , speed , query source | claude code.api retries exhausted | All retry attempts are exhausted | model , error , total attempts , total retry duration ms , status code | claude code.tool result | Tool execution completes | tool name , success , duration ms , decision type , decision source , error type , tool input size bytes , tool result size bytes | claude code.tool decision | Tool permission is accepted or rejected | tool name , decision accept / reject , source config , hook , user permanent , user temporary , user abort , user reject | claude code.permission mode changed | Permission mode switches | from mode , to mode , trigger shift tab , exit plan mode , auto gate denied , auto opt in | claude code.mcp server connection | MCP server connects, fails, or disconnects | status connected , failed , disconnected , transport type , server scope , duration ms , error code | claude code.auth | /login or /logout completes | action login / logout , success , auth method , error category | claude code.compaction | Conversation compaction completes | trigger auto / manual , success , pre tokens , post tokens , duration ms | claude code.internal error | Unexpected internal error is caught | error name , error code | claude code.plugin installed | Plugin finishes installing | plugin.name , marketplace.name , install.trigger | claude code.plugin loaded | Plugin is active at session start | plugin.name , plugin.scope , plugin.version , enabled via | claude code.skill activated | Skill is invoked by Claude or as a / command | skill.name , invocation trigger , skill.source | claude code.hook execution complete | All hooks for a hook event finish | hook event , hook name , num hooks , num success , num blocking , total duration ms | claude code.feedback survey | Session quality survey is shown or answered | event type , survey type , response | All events share a prompt.id attribute, a UUID that links every event produced while processing a single user prompt. Filter by prompt.id in SigNoz to correlate all API requests and tool executions triggered by one prompt. Troubleshooting If you don't see your telemetry data in SigNoz: Check - This must be set to CLAUDE CODE ENABLE TELEMETRY 1 . Telemetry is opt-in and nothing is exported without it Verify your region and ingestion key - An incorrect <region or <your-ingestion-key will silently drop all data. Double-check against your SigNoz Cloud settings https://signoz.io/docs/ingestion/signoz-cloud/keys/ Check both exporters are set - OTEL METRICS EXPORTER controls metrics and OTEL LOGS EXPORTER controls events. If only one is set, the other signal will not appear Try a console exporter - Set OTEL METRICS EXPORTER=console and OTEL LOGS EXPORTER=console to confirm Claude Code is generating telemetry locally before sending to SigNoz Wait for the export interval - The default metrics export interval is 60 seconds. During setup, set OTEL METRIC EXPORT INTERVAL=10000 to flush every 10 seconds instead Confirm env vars are set before launch - Environment variables must be set before launching claude or code . . Variables exported after launch have no effect FAQs Does Claude Code have built-in monitoring? A. Yes. Claude Code has native OpenTelemetry support that you opt into by setting CLAUDE CODE ENABLE TELEMETRY=1 . Once enabled, it emits metrics token usage, costs, session counts, lines of code, commits and structured log events API requests, tool executions, permission decisions via the standard OTLP protocol to any compatible backend including SigNoz. How do I track Claude Code token usage? A. Claude Code emits a claude code.token.usage metric after every API request, broken down by type input , output , cacheRead , cacheCreation , model , and query source . In SigNoz, filter this metric by user or model to see exactly where tokens are going. See the Telemetry Data telemetry-data section for the full attribute list. Can I monitor Claude Code costs with OpenTelemetry? A. Yes. The claude code.cost.usage metric tracks estimated USD cost per API request and includes attributes for model , agent.name , skill.name , and query source , so you can break down spend by team, model, or workflow. Note that these are approximations — for billing data, refer to your Anthropic Console or cloud provider. How do I monitor Claude Code across an entire team? A. Use the Administrator Configuration section to deploy a managed settings file managed-settings.json via MDM or your device management solution. This sets CLAUDE CODE ENABLE TELEMETRY=1 and the SigNoz endpoint centrally for all users, without requiring each developer to configure env vars individually. What is the difference between Claude Code metrics and events? A. Metrics like claude code.token.usage and claude code.cost.usage are numeric time-series data, ideal for dashboards, aggregations, and alerts. Events like claude code.api request and claude code.tool result are structured log records emitted per action, ideal for audit trails, debugging, and tracing individual sessions. Both are exported via OpenTelemetry but require separate exporter configuration OTEL METRICS EXPORTER vs OTEL LOGS EXPORTER . Related Articles Bringing Observability to Claude Code: OpenTelemetry in Action https://signoz.io/blog/claude-code-monitoring-with-opentelemetry/ A practical walkthrough of how to connect Claude Code's monitoring hooks with OpenTelemetry and visualize usage, costs, and performance in SigNoz dashboards. Is Claude Code Getting Worse? How to Measure Degradation with OpenTelemetry https://signoz.io/blog/claude-code-measure-degradation-opentelemetry/ Learn how to track output-per-token ratios across commits, PRs, and lines of code to detect whether Claude Code's efficiency is degrading over time.