EzFlowEzFlow Docs
Messaging & BYOK

LLM API keys

Adding your own OpenAI, Anthropic, or Gemini API keys for AI nodes.

EzFlow's AI nodes use bring-your-own keys — your OpenAI, Anthropic, or Gemini API key is stored encrypted in your workspace and decrypted only at execution time. Keys are never stored in Workflow DAGs or visible in run logs.

Supported providers

ProviderProvider value used in nodes
OpenAIopenai
Anthropicanthropic
Geminigemini

Each key you add is stored as an Integration record encrypted with AES-256-GCM. You then reference the integration by its ID when configuring an LLM_COMPLETION node.

Adding an LLM key

  1. Go to Integrations (/integrations).
  2. Click New integration.
  3. Set the Provider to openai, anthropic, or gemini.
  4. Paste your API key as the access token.
  5. Give it a descriptive name (e.g. OpenAI production key) and save.

You can add multiple keys per provider — for example, separate keys for development and production, or different keys per team.

Using an LLM key in a Workflow

  1. Add an LLM_COMPLETION node to your Workflow.
  2. Set provider to match the key you added (openai, anthropic, or gemini).
  3. Select the integration you created from the integrationId field.
  4. Choose a model (e.g. gpt-4o, claude-3-5-sonnet-latest, gemini-1.5-pro) and configure your prompts.

The node decrypts the key at run time and calls the provider's API. The raw key never appears in the Workflow DAG or run logs.

  • LLM_COMPLETION — the AI node that uses the key to call a model
  • Integrations — where LLM keys and other credentials are stored

Sending domains

Verifying custom domains for email deliverability and branding.

Integrations

Connecting third-party services and external APIs to EzFlow workflows.

On this page

Supported providersAdding an LLM keyUsing an LLM key in a WorkflowRelated