Back to Skills

arize-ai-provider-integration

Creates, reads, updates, and deletes Arize AI integrations that store LLM provider credentials used by evaluators and other Arize features. Supports any LLM provider (e.g. OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Vertex AI, Gemini, NVIDIA NIM). Use when the user mentions AI integration, LLM provider credentials, create integration, list integrations, update credentials, delete integration, or connecting an LLM provider to Arize.

38stars5forksUpdated 7/17/2026

Security Assessment

Critical Risk(15/100)

Detected risks:

Persistence([references/ax-profiles.md] .zshrc, [references/ax-profiles.md] .bashrc)
Secret Exposure([references/ax-profiles.md] API_KEY=)
Sensitive File Access([SKILL.md] .env, [references/ax-profiles.md] .Env)
Security Score15/100

About arize-ai-provider-integration

This skill creates, reads, updates, and deletes Arize AI integrations that store LLM provider credentials used by evaluators and other Arize features that need to invoke an LLM. It solves the problem of registering and managing provider credentials in Arize in a consistent way, supporting many providers including OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Vertex AI, Gemini, NVIDIA NIM, and OpenAI-compatible custom endpoints. An AI integration is stored provider credentials referenced by a base64 integration ID, which downstream operations such as evaluator creation require.

All operations run through the ax CLI. You can list integrations with ax ai-integrations list --space SPACE, including name filtering, pagination via --limit and --cursor, and JSON output; get details with ax ai-integrations get NAME_OR_ID; and create with ax ai-integrations create, where required flags depend on the provider. API-key providers like openAI, anthropic, and gemini use --api-key; azureOpenAI and nvidiaNim also need --base-url; awsBedrock uses --provider-metadata with a role_arn; and vertexAI uses --provider-metadata with project_id and location. Optional flags include --model-name (repeatable), --enable-default-models, and --function-calling-enabled. The skill notes that create is account-scoped and does not accept --space, while list, get, update, and delete do. The SPACE value can be a space name or a base64 space ID, discoverable via ax spaces list, and auth types include default, proxy_with_headers, and bearer_token.

The skill targets Arize users connecting LLM providers for evaluators and other features, and it requires the ax CLI and a configured Arize profile. It emphasizes strong security practices: never read .env files or search the filesystem for credentials, never ask users to paste secrets into chat, and reference env var names like $OPENAI_API_KEY rather than raw values, asking the user to export keys and run create in their own terminal. Troubleshooting guidance maps common errors, such as command not found, 401 Unauthorized, or unknown space, to reference docs like references/ax-setup.md and references/ax-profiles.md, and directs users to https://app.arize.com/admin for API keys.

FAQ

What are the prerequisites?

The ax CLI and a configured Arize profile. You proceed directly by running the ax command you need, and troubleshoot based on any error rather than checking versions or profiles upfront.

Which LLM providers are supported?

Any LLM provider, including openAI, anthropic, azureOpenAI, awsBedrock, vertexAI, gemini, nvidiaNim, and custom OpenAI-compatible endpoints. Required flags vary: API-key providers use --api-key, Azure and NVIDIA NIM also need --base-url, and Bedrock and Vertex use --provider-metadata.

Does creating an integration require a space?

No. ai-integrations create does not accept --space because AI integrations are account-scoped. Use --space only with list, get, update, and delete.

How does the skill handle credentials securely?

It never reads .env files or searches the filesystem for credentials and never asks users to paste secrets into chat. It references env var names like $OPENAI_API_KEY and asks the user to export the key and run the create command in their own terminal.

What do I do about a 401 Unauthorized error?

Run ax profiles show to inspect the current profile, and if it is missing or the API key is wrong, follow references/ax-profiles.md to create or update it. If the user lacks a key, direct them to https://app.arize.com/admin > API Keys.

All Files

3 files
SKILL.md11.1 KB
View
references/ax-profiles.md4.5 KB
View
references/ax-setup.md1.5 KB
View

Install arize-ai-provider-integration

Download and extract the skill files to your .claude/skills/ directory.

Quick Setup:

  1. Copy the skill folder to .claude/skills/
  2. Claude will automatically detect and use the skill