Quickstart
This quickstart walks you through sending trace data to the Observability Stack and viewing it in OpenSearch Dashboards.
Prerequisites
Section titled “Prerequisites”Make sure the stack is running. If you haven’t installed it yet, see Get Started.
Step 1: Send traces
Section titled “Step 1: Send traces”The fastest way to send traces is with the Python SDK:
pip install opensearch-genai-sdk-pyfrom opensearch_genai_sdk_py import register, workflow, tool
register(endpoint="http://localhost:4318/v1/traces", service_name="my-app")
@tool(name="process_data")def process_data(size: int) -> dict: return {"processed": size}
@workflow(name="my-first-workflow")def run(): return process_data(42)
run()For other languages, see Send Data.
Step 2: View traces
Section titled “Step 2: View traces”- Open OpenSearch Dashboards at
http://localhost:5601. - Navigate to Observability > Traces.
- You should see your trace appear within a few seconds.
Step 3: Explore the dashboard
Section titled “Step 3: Explore the dashboard”From the Dashboards home, explore:
- Traces: Search and filter trace data with PPL queries.
- Services: View the auto-generated service map.
- Metrics: Query Prometheus metrics with PromQL.
Next steps
Section titled “Next steps”- Ingest Your First Traces — detailed trace instrumentation guide
- Create Your First Dashboard — build custom visualizations
- Agent Tracing — trace AI agent workflows