This endpoint allows clients to send prompts to a specified agent and receive responses. It supports both streaming and non-streaming responses. Visit fullmetal to get {{agentId}} and {{apikey}}.
\nRequest Headers
\nRequest Body
\nThe request body should be in JSON format and must include the following fields
\n{\n \"agentId\": \"string\", // The ID of the agent to which the prompt is sent (required)\n \"prompt\": \"string\", // The prompt text to be sent to the agent (required)\n \"systemPrompt\": \"string\", // Optional system prompt to provide context (optional)\n \"stream\": \"boolean\" // Indicates whether to use streaming for the response (optional)\n}\n\nThe response will be in JSON format and can vary based on whether streaming is enabled or not.
\nFor non-streaming responses:
\n{\n \"response\": \"The capital of France is Paris.\",\n \"model\": \"model_name\", // The model used to generate the response\n \"speed\": \"number\", // The speed of the response\n \"elapsedTime\": \"number\", // The time taken to generate the response\n \"reasoning\": \"string\", // Reasoning behind the response (if applicable)\n \"summarizedText\": \"string\" // Summarized text (if applicable)\n}\n\nFor streaming responses: The response will be sent in chunks, and each chunk will contain:
\n{\n \"token\": \"string\", // The generated token (part of the response)\n \"completed\": \"boolean\", // Indicates if the response is complete\n \"model\": \"model_name\", // The model used to generate the response\n \"speed\": \"number\", // The speed of the response\n \"elapsedTime\": \"number\" // The time taken to generate the response\n}\n\nIn case of an error, the response will include:
\n{\n \"type\": \"error_type\", // Type of the error\n \"message\": \"error_message\", // Description of the error\n \"status\": \"number\" // HTTP status code\n}\n\n