OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
MCP & Tooling

Opik MCP Server

Opik MCP Server is a Python Model Context Protocol server that connects compatible AI clients to an Opik workspace for retrieving traces, logging scores and saving prompt versions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Opik MCP Server — OpenViST project profile in MCP & Tooling.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when an MCP-compatible AI client should interact with an existing Opik workspace as part of an engineering conversation.

Who it is for

  • LLM engineers who already use Opik and want to work with its traces, scores or prompts through an AI assistant.
  • Teams using an MCP-compatible client with Opik Cloud, self-hosted Comet or open-source Opik.

Documented uses

  • Ask an AI client to retrieve traces from an Opik project, such as finding failed traces.
  • Record an evaluation score and rationale against a trace.
  • Save prompt versions from an assistant conversation.
Before you choose

Requirements & limits

The Opik Cloud hosted server uses browser sign-in and requires no API key according to the README. The local server uses environment variables; an API key is required only where the deployment requires one. The repository specifies the Apache-2.0 license.

  • Opik Cloud uses Comet's hosted Streamable HTTP endpoint; self-hosted Comet and open-source Opik use the local package over stdio.
  • The uv-based setup command registers the server, installs the skill pack and checks the connection; manual setup skips those extra steps.
  • For self-hosted Comet with named workspaces, configure OPIK_WORKSPACE to avoid reads defaulting to the account's default workspace.
  • Open-source Opik exposes its API at /api and uses a single default workspace.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Configure Opik MCP for a supported AI client

Register the MCP server, install the Opik skill pack and check the connection using the documented setup command.

Before you start

  • The uv tool.
  • An Opik deployment saved in ~/.opik.config, available on localhost, or a deployment to select when prompted.
  • A supported client: Claude Code, Codex, Cursor, VS Code Copilot or opencode.

Steps

  1. Run uvx opik@latest mcp configure.

  2. If configuring from a coding agent or script, specify the client, for example uvx opik@latest mcp configure --ai-client claude-code --skills.

  3. If prompted, identify the Opik deployment to use.

  4. Start a new client session so it loads the MCP server.

Expected outcome

The client is configured to connect to Opik; the setup command also installs the skill pack and checks the connection.

Guide limitations

  • For Opik Cloud, the client signs in through a browser; no API key is required for the hosted server.
  • The command-line flow requires uv.
  • Manual server registration skips the deployment check and skill-pack installation.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
comet-ml/opik-mcp · Official homepage title · retained in its source language
How we review sourcesSuggest a correction