MCP For Claude Marketplace
Arize AI

Arize AI MCP Connector for Claude

A+

Automate LLM and ML observability via Arize — monitor models, track telemetry, run evaluations, and analyze data drift directly from any AI agent.

10 tools Official Updated Jun 28, 2026 Official Vinkius Partner
More Details Connect to Claude

Connect your Arize AI observability platform to any AI agent and take full control of your Machine Learning and LLM telemetry workflows through natural conversation.

What you can do

  • Model Monitoring & Metrics — List all tracked ML models, extract deep configuration schemas, and fetch real-time metrics (performance, data quality, and prediction drift)
  • Evaluation & Alignment — Launch and list automated LLM evaluation runs (e.g., Toxicity, Hallucination, PII filtering) against static datasets and ground truth baselines
  • Telemetry Ingestion — Push programmatic raw logs, predictions, and inferences straight into Arize for immediate visualization and tracking
  • Space & Environment Management — Browse organizational spaces and segregated deployment environments (Production, Training, Verification)

How it works

  1. Subscribe to this server
  2. Enter your Arize API Key and Space ID Key
  3. Start monitoring your prediction health from Claude, Cursor, or any MCP-compatible client

No more context-switching into heavily graphical dashboards to figure out why an LLM prompt hallucinated. Your AI acts as a dedicated ML Ops engineer.

Who is this for?

  • Machine Learning Engineers — rapidly push inference telemetry and query performance degradation flags without leaving your terminal
  • AI Product Managers — instantly monitor output toxicity, drift rates, and usage metrics across multiple LLM integrations
  • Data Scientists — manage baseline evaluation datasets and trigger custom scoring loops asynchronously
ml-observabilityllm-evaluationmodel-monitoringtelemetrydata-driftai-alignment

10 tools expose this connector's capabilities to your AI agent.

get_dataset

Get a specific evaluation dataset

get_model

It defines the inputs, outputs, and features. Get details and metadata for a specific tracked model

list_spaces

Spaces separate different models and telemetry datasets. List accessible workspaces within the Arize platform

list_datasets

List static evaluation datasets

list_environments

g., Production, Training, Verification) used to segregate model inferences and baseline datasets. List configured environments within Arize

list_evals

g., Toxicity, Hallucination, PII filtering). List automated evaluation runs

ingest_log

payload_json must contain valid Arize payload structures. Ingest raw telemetry logs into Arize

get_metrics

Fetch observability metrics for an ML model

list_models

List tracked ML models or LLMs

run_eval

Trigger a custom LLM evaluation run

See how to talk to your AI agent using Arize AI.

List all active Machine Learning models monitored in my workspace.

I've fetched your models from the Arize Space. You currently have 3 connected models: 'Fraud-Detection-v2' (Classification), 'Customer-Churn-XGB' (Score), and 'OpenAI-Customer-Service-Bot' (LLM). Would you like to see the recent drift metrics for any of them?

Get the evaluation baseline datasets available for our LLM checks.

You have two static evaluation datasets loaded in Arize: 'Support-Tickets-Q2-GroundTruth' (1400 rows) and 'Toxicity-Benchmark' (250 rows). I can trigger an automated `run_eval` check targeting these datasets against your active LLM logs if needed.

Push these 3 mocked prompt responses as telemetry logs to the 'OpenAI-Customer-Service-Bot' model.

I successfully structured your 3 prompts into valid ingestion payloads and pushed them via the `ingest_log` tool. They should now be available for analysis and drift observation in the Arize telemetry dashboard.

Yes! You can ask your agent to retrieve the specific Ground Truth dataset ID, formulate a testing payload, and invoke the `run_eval` tool natively. Arize will process the asynchronous scoring internally and log the evaluation securely.

Related Connectors

SS

Salt Security MCP

10 tools

Integrate Salt Security directly with your AI for comprehensive API threat vector discovery, posture management, and active remediation in real-time.

A+ View details →
RI

Reddit Intelligence MCP

9 tools Official

Tap into Reddit's public communities for real-time marketing intelligence — discover trending topics, monitor niches, and analyze sentiment with a free Reddit app.

A+ View details →
OE

Open Exchange Rates MCP

7 tools Official

Access real-time and historical currency exchange rates for 200+ currencies. Convert values, track time series, and analyze OHLC data directly from your AI agent.

A+ View details →
ID

Inform Direct MCP

10 tools Official

File UK company documents with Companies House digitally and manage statutory records, share registers, and annual filings.

A+ View details →