MCP Servers

Arize AI's Phoenix, an open-source AI observability and evaluation platform, exposes a remote MCP endpoint that connects agents directly to a Phoenix instance. Lets assistants query OpenTelemetry traces and spans from LLM applications, inspect datasets and experiments, review evaluation results, and investigate latency or quality regressions.

The official Weights & Biases MCP server for W&B Models and Weave, letting AI assistants query and analyze experiment tracking data in natural language. Agents can look up runs and their metrics, compare sweeps, inspect registered model artifacts, and dig into Weave traces to debug LLM application behavior.

Confident AI's official MCP server, connecting AI tools to the platform behind DeepEval for evaluating, observing, and iterating on LLM quality. Agents can trigger and read evaluation runs, inspect metric scores and failing test cases, manage golden datasets, and review production traces to close the loop between eval results and code changes.

Comet's official MCP server for Opik, an open-source LLM evaluation and observability platform. Connects an AI host directly to an Opik workspace so agents can read traces and spans, log evaluation scores, manage prompt versions in the prompt library, and inspect project metrics without leaving the editor.