InfraA capability for your stack
Observability and evaluation
Inspect agent behavior and evaluate application quality.
The directory
Tools for the job
Listed A–Z. Hosted and self-hosted options share one listing. About the labels
AgentOps
AgentOpsAgent observability software for recording runs, inspecting execution graphs, and monitoring application activity and model usage.
Details & sources
The repository includes a self-hosted dashboard and API backend, alongside the SDK.
Source repository ↗ · AgentOps repository including the app and backend.
Reviewed
Arize Phoenix
Arize AISelf-hosted observability and evaluation software for tracing agent applications and running experiments with datasets.
Self-hostedSource availableElastic-2.0Details & sources
The reviewed Phoenix runtime uses Elastic License 2.0. Arize AX is a separate managed product.
Source repository ↗ · Phoenix application runtime; some related packages may have different licenses.
Reviewed
Braintrust
BraintrustAn agent observability platform for collecting traces, exploring production behavior, and evaluating changes against datasets.
HostedSelf-hosted data planeDetails & sources
Enterprise self-hosted data plane. The UI, authentication, and control plane remain managed by Braintrust.
Reviewed
DeepEval
Confident AIAn evaluation framework for running tests against agent outputs, trajectories, tool use, and individual application components.
Self-hostedOpen sourceApache-2.0Details & sources
The framework runs in your environment. Confident AI is a separate optional hosted platform; model-based evaluators may call configured external providers.
Source repository ↗ · DeepEval evaluation framework; optional Confident AI service terms are separate.
Reviewed
Langfuse
LangfuseAgent observability and evaluation software for inspecting traces, working with datasets, and testing application behavior.
Details & sources
Community code is MIT-licensed; enterprise directories and some add-on features require separate licensing.
Source repository ↗ · Community code outside ee/, web/src/ee/, and worker/src/ee/. Enterprise code has a separate license.
Reviewed
LangSmith
LangChainAn observability and evaluation platform for tracing agent runs, monitoring production behavior, and working with evaluation datasets.
HostedSelf-hostedDetails & sources
Enterprise plan add-on requiring a license key; the platform runs in the customer infrastructure.
Reviewed
No tools match those filters.
Try a different search or see all deployment options.
Know a tool that belongs here? Send it our way ↗