Documentation
Viz Cloud SDK & API Reference
Instrument agents, configure the telemetry pipeline, and operate agents in production.
Instrument your first agent in under 5 minutes.
- Install the SDK
- Configure Viz Cloud
- Your first instrumented agent
- View your data
Full API for the Viz Cloud Python SDK.
- routeiq.configure()
- @routeiq.instrument
- routeiq.session()
- sess.tool()
- sess.guardrail()
- sess.escalate()
Framework Integrations
Drop-in instrumentation for popular agent frameworks.
- LangGraph
- OpenAI Agents SDK
- CrewAI
- LangChain
- LlamaIndex
- Custom runtimes
OTel Collector Setup
Configure the OpenTelemetry pipeline for your environment.
- Local collector
- Datalake exporter
- Environment variables
- Cloud / managed setup
Span schemas, attribute names, and metric definitions.
- Span types
- Log attributes
- Metric names
- Data model
Guides
Step-by-step guides for common production patterns.
- Setting up reliability SLOs
- Debugging a stuck run
- Multi-agent topology
- Version canary setup
Telemetry Taxonomy
Data model
| Signal | Contents |
|---|---|
| otel_logs | One row per event: api_request, tool_decision, tool_result, user_prompt |
| otel_metrics_sum | Token counts, cost (cost_usd), active time — aggregated by metric name |
| otel_traces | One span per tool call (tool.call) and per session (agent.session) |
Need help getting started?
Viz Cloud is in private beta. Reach out and we'll help you instrument your specific agent stack.
Request early access