IdentArk icon

IdentArk

Route LLM calls through the IdentArk gateway so your workflows hold zero API keys

Actions9

Session → Create

AI-generated

Summary

Create a credential-isolated session for an agent through the IdentArk gateway. The session uses the selected provider and model, references an LLM credential stored in the vault, and can reject calls after reaching its USD cost cap.

Inputs

  • Agent ID (required) — Identifier of the agent that owns the session.
  • Model — LLM model to use; defaults to GPT-4o. Options include Claude Haiku, Claude Sonnet 4, GPT-4o, GPT-4o Mini, Mistral Large, and Mistral Small.
  • Provider — Provider to route the session through; defaults to OpenAI. Options include Anthropic, AWS Bedrock (UK), Azure OpenAI (UK), Mistral AI (EU), and OpenAI.
  • Credential Reference (required) — Vault path to the provider credential, such as secret/orgs/{org}/providers/openai.
  • Cost Cap (USD) — Maximum session spend before calls are rejected; defaults to 5 USD.

Output shape

One output item per input item, containing the JSON response from the session creation request.

Sends an authenticated POST request to /v1/sessions, mapping the inputs to agent_id, model, provider, credential_ref, and cost_cap_usd. Errors are raised unless the node is configured to continue on fail, in which case the affected item contains an error field.

Links

Discussion