Amazon Bedrock Agents

Amazon Bedrock Agents are AWS-managed services that combine foundation models with external data sources and APIs to create deployable conversational agents. Muster captures Bedrock Agent runs through OpenTelemetry instrumentation — a vendor-neutral path that works regardless of how you deploy the agent.

Setup

The recommended path follows the AWS-maintained langfuse-genaiops notebook, which contains complete configuration examples for Bedrock Agents with Muster.

The high-level steps:

1. Configure Muster credentials as environment variables (LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY, LANGFUSE_BASE_URL).
2. Set up OpenTelemetry to export to Muster's /api/public/otel endpoint (see the OpenTelemetry integration).
3. Instrument your Bedrock Agent invocation code so every agent step, tool call, and model completion produces a span.
4. View the trace in Muster, including agent reasoning, knowledge base retrievals, action group invocations, and token usage.

See also

• Upstream Langfuse Bedrock Agents docs
• Amazon Bedrock
• OpenTelemetry