Honeycomb + Amazon Bedrock AgentCore
Honeycomb's production integration with Amazon Bedrock AgentCore is here, surfacing agent telemetry directly in Agent Timeline, our trace view for AI agent behavior. Available today and built on OpenTelemetry.
Full-stack visibility into your AgentCore agents
The Honeycomb integration connects your Bedrock AgentCore agents to Agent Timeline, giving you a unified trace view of what your agent did, what changed in your environment, and how behavior evolved over time.
Agent Timeline
See every agent action in a single trace view, correlated with deployments and config changes.
LLM-specific traces
GenAI semantic conventions surface model calls, token counts, and reasoning steps natively.
Dataset routing
Route traces to the right Honeycomb dataset using x-honeycomb-dataset headers.
Step-by-step guide
A Jupyter notebook in the official Amazon Bedrock AgentCore samples repo walks you through setup end-to-end.
Built on OpenTelemetry without lock-in
If your team already uses OpenTelemetry for your services, this integration extends naturally with the same instrumentation and pipeline. If you're starting from scratch, you're building on a foundation that's portable and vendor-neutral.
Canvas is built on AgentCore
Honeycomb's AI workspace for production debugging starts with AgentCore and ends with features that help your team investigate with accuracy.
Session isolation
Every Canvas session runs in its own context. No state bleeding between users, no custom isolation layer to build.
Code interpreter
Canvas generates and executes queries inside a secure, managed AgentCore sandbox.
Memory management
AgentCore handles what gets stored and retrieved across a multi-turn investigation without manual history re-injection.
Official Amazon AgentCore partner
The Honeycomb integration is included in the official AWS agentcore-samples repository as a partner observability example, a reference for teams building on AgentCore.