UniCraft AI Agent

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

Actions2

Overview

The UniCraft AI Agent node enables interaction with an AI agent that supports conversational memory, tool integrations, and smart model routing. It is designed to facilitate advanced AI-driven conversations where the agent can remember context, call external tools, or dynamically select the best AI model based on a strategy.

This node is beneficial in scenarios such as:

  • Building chatbots that maintain conversation history for more coherent interactions.
  • Automating workflows where the AI agent can invoke external APIs or functions (tools) to enrich responses.
  • Optimizing AI usage by automatically selecting models based on cost, performance, or availability preferences.

For example, you could use this node to create a customer support assistant that remembers previous messages, calls a product database API to fetch details, and chooses the most appropriate AI model depending on current load or cost constraints.

Properties

Name Meaning
Operation Selects the operation mode: "Chat" to interact normally with the AI agent, or "Manual Trigger" to manually send input data to the agent.
Agent Type Type of AI agent to create. Options include:
- Conversational Agent
- Tool-Enabled Agent
- Smart Routing Agent
- Custom Agent
Input Message The message text sent to the AI agent when using Manual Trigger operation.
System Message Defines the agent's behavior and personality; acts as a system prompt guiding the AI's responses.
Memory Configuration Configures how the agent maintains conversation memory. Options:
- Memory Type (buffer, summary, token_buffer, window, none)
- Max Messages (for window type)
- Max Tokens (for token_buffer type)
- Summary Threshold (for summary type)
Model Configuration Configures the AI model used by the agent:
- Mode: Smart Routing (auto-select model) or Manual (select specific model)
- Model (when manual)
- Smart Routing Strategy (balanced, cost optimized, etc.)
- Temperature (sampling randomness)
- Max Tokens (max tokens to generate)
Tools Configuration Configure tools available to the agent:
- Enable Tools (boolean)
- Tool list: each with Name, Description, Parameters (JSON schema), and API Endpoint
Advanced Configuration Advanced options:
- Max Iterations (for tool-enabled agents)
- Timeout (seconds) for agent execution
- Enable Streaming (boolean)
- Return Intermediate Steps (boolean)

Output

The node outputs JSON objects containing the AI agent's response and metadata:

  • response: Object with
    • role: Role of the responder (usually "assistant")
    • content: Text content of the AI's reply
    • tool_calls: Array of any tool invocations made during the response
  • usage: Token usage statistics from the AI provider
  • model: The AI model used for generating the response
  • provider: Fixed string "unicraft"
  • agentType: The selected agent type
  • text / message: The textual response repeated for convenience
  • conversationId: Identifier for the conversation session
  • memory: If memory is enabled, includes memory type and configuration
  • smartRouting: Present if smart routing mode is used, showing strategy and selected model
  • tools: Information about tools available and called during the response
  • intermediateSteps: Optional detailed info about request/response and processing time if enabled

No binary data output is produced by this node.

Dependencies

  • Requires an API key credential for the UniCraft API service.
  • Needs proper configuration of the UniCraft API base URL and API key in credentials.
  • Uses UniCraft API endpoints for:
    • Fetching available AI models
    • Fetching smart routing strategies
    • Sending chat completion requests (either standard or smart routing)
  • Network connectivity to UniCraft API is required.

Troubleshooting

  • Missing Credentials: The node will throw an error if the UniCraft API base URL or API key is not set or invalid. Ensure credentials are configured correctly.
  • Empty User Message: If no valid input message is found in the input data or provided manually, the node errors out. Provide a non-empty message via input or parameters.
  • Invalid Input Data Format: The node expects input JSON to contain one of these fields for the user message: message, text, prompt, query, or a messages array. Incorrect formats cause errors.
  • API Request Failures: Network issues or invalid API keys will result in request errors. Check API key validity and network access.
  • Tool Configuration Errors: If tools are enabled but improperly configured (e.g., missing required fields), the node may fail or behave unexpectedly.
  • Timeouts: Long-running requests may timeout based on the configured timeout setting; increase it if needed.
  • Continue On Fail: When enabled, the node returns error information per item instead of stopping execution.

Links and References

Discussion