UniCraft AI Agent

UniCraft AI Agent with Memory, Tools, and Smart Routing capabilities

Actions2

Manual Trigger

AI-generated

Summary

Send a manual input message to a configurable UniCraft AI agent to receive a generated AI response with optional memory, model, multi-modal, and tool integrations.

Inputs

  • Agent Type (required) — Select the type of AI agent to create, such as conversational, tool-enabled, smart routing, or custom.
  • Input Message (required) — The text message to send to the AI agent for processing.
  • System Message — A system prompt defining the agent's behavior and personality.
  • Memory Configuration — Settings to configure conversation memory behavior, including memory type, maximum messages or tokens retained, or summarization thresholds.
  • Model Configuration — Options to configure the AI model selection mode (smart routing or manual), specific model, routing strategy, generation temperature, and max tokens to generate.
  • Multi-Modal Configuration — Settings for input and output types to handle text, images, audio, video, documents, code, or markdown along with optional media file uploads for multi-modal processing.
  • Tools Configuration — Enable and define external tools (functions) that the agent can call during execution, specifying their name, description, parameters, and API endpoints.
  • Advanced Configuration — Advanced settings including maximum iterations for tool-enabled agents, execution timeout, streaming response enablement, and returning intermediate execution steps.

Output shape

a single JSON object containing the AI agent's response message, tool call data if any, usage and cost metadata, optional media data for multi-modal outputs, conversation memory metadata, and optionally intermediate execution steps.

The response includes the agent's reply text under 'response.content' and may include additional fields for tool calls, memory configuration used, smart routing details, and any generated media files. If streaming is enabled or intermediate steps requested, those data appear within the output JSON. Errors per input item surface only if 'Continue On Fail' is enabled.

Examples

Example 1: Use a conversational AI agent to process a custom user input message and receive a generated text reply.

Set Agent Type to 'Conversational', enter the Input Message you want the agent to respond to, optionally customize the system prompt and memory settings, and run the operation.

Discussion