iSharkFly-Docs/amazon-bedrock-agentcore-samples

mirror of synced 2026-05-22 22:53:35 +00:00

Files

T

Bharathi Srinivasan e746bf7764 Features folder revamp (#1540 )

adding scripts for agentcore features; jupyter notebooks moved to workshops; reorganising folders

2026-05-20 18:35:16 -07:00

1.8 KiB

Raw Permalink Blame History

AgentCore + Datadog LLM observability

Deploy a Strands travel agent to AgentCore runtime with traces sent to Datadog LLM observability via OTLP HTTP.

Architecture

AgentCore runtime → travel_agent.py
  └── Custom OTelTracerProvider
        └── OTLPSpanExporter → https://trace.agent.{DD_SITE}/v1/traces
              headers: dd-api-key, dd-otlp-source=llmobs
                └── Datadog LLM observability dashboard

DISABLE_ADOT_OBSERVABILITY=true bypasses the default CloudWatch ADOT pipeline. OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental enables OTel v1.37+ GenAI semantic conventions required for Datadog LLM observability views.

Prerequisites

Python 3.10+, uv
AWS credentials configured
Datadog account with API key

Quick Start

pip install bedrock-agentcore boto3 python-dotenv
cp .env.example .env
# Edit .env: set DD_API_KEY (and optionally DD_SITE for non-US1 regions)
python deploy.py
python invoke.py
# View traces: https://app.datadoghq.com/llm/traces
python cleanup.py

Datadog Regions

Region	DD_SITE
US1 (default)	`datadoghq.com`
US3	`us3.datadoghq.com`
US5	`us5.datadoghq.com`
EU1	`datadoghq.eu`
AP1	`ap1.datadoghq.com`

Files

File	Description
`utils/travel_agent.py`	Agent with Datadog OTel TracerProvider
`deploy.py`	Deploys to AgentCore runtime with Datadog env vars
`invoke.py`	Invokes the deployed agent
`cleanup.py`	Deletes all created AWS resources

1.8 KiB Raw Permalink Blame History