DevanaInstall

Summary

Send a chat completion request to the Devana AI API using a specified agent to generate responses based on input messages and optional parameters.

Inputs

• Agent ID (Model) (required) — Select the Devana agent ID to use for generating the chat completion, either from a searchable list or by directly providing the ID.
• Messages (required) — Provide one or more messages to send to the model, each with a role (system, user, or assistant) and content string.
• Additional Fields — Optional parameters to customize the chat completion, including temperature, max tokens, top P, streaming option, conversation ID to continue, language, response format (text or JSON object), included files, frequency and presence penalties, custom identity, hidden mode, stop sequences, and custom metadata as JSON.

Output shape

Two output streams: the first outputs the chat completion response messages per input item; the second outputs metadata including usage statistics, finish reason, model info, possible sources, and score for each item.

The chat response is contained in the first output with its message under the 'message' field. If 'response_format' is set to 'json_object', the message content is parsed as JSON if possible. Additional metadata for each response is included in the second output stream. Errors appear in the first output under an 'error' field if continueOnFail is enabled.