AI Agent

ai.agent AI v0.1.0

Conversational AI agent with a pluggable model (Ollama local/cloud), pluggable memory (state/file/in-memory), and tools (workflows + MCP). Calls tools in a loop until it can answer.

Ports

Direction	Port	Label
Input	`input`	Message
Output	`output`	Reply

Properties

Property	Type	Default	Description
Model provider `provider`	`select`	`"ollama"`	Where the model runs. 'ollama' = a local Ollama daemon; 'ollama-cloud' = ollama.com (needs an API key). `ollamaollama-cloud`
Model `model`	`string`	`"qwen2.5:0.5b"`	Model name, e.g. qwen2.5:0.5b (local) or gpt-oss:20b (cloud).
Base URL `baseUrl`	`string`	`""`	Override the endpoint. Defaults to http://localhost:11434 (local) or https://ollama.com (cloud).
API key `apiKey`	`string`	`""`	Bearer key for the cloud provider. Leave blank for a local daemon.
System prompt `system`	`string`	`"You are a helpful assistant."`	The agent's persona and instructions.
Rules `rules`	`json`	`[]`	A list of short rules the agent must always follow.
Tools `tools`	`json`	`[]`	Tools the agent may call. Each: { type:'workflow', name, description, parameters, workflowId } or { type:'mcp', command, args }.
Output format `outputFormat`	`select`	`"text"`	text = a free-form reply; json = constrain the final answer to a JSON object. `textjson`
Memory `memoryBackend`	`select`	`"state"`	Where conversation history lives. state = durable host store; file = flat JSON files; inmemory = ephemeral. `statefileinmemory`
Session key `sessionKey`	`string`	`"{{ input.sessionId }}"`	Template for the conversation id, e.g. {{ input.sessionId }}. Same key = same memory.
User message `message`	`string`	`"{{ input.message }}"`	Template for the user's message, e.g. {{ input.message }} or {{ input }}.
Temperature `temperature`	`auto`	`0.2`	Sampling temperature (0 = deterministic, 1 = creative).
Max tokens `maxTokens`	`int`	`512`	Max tokens the model may emit per turn.
Max tool rounds `maxIterations`	`int`	`6`	How many tool-call rounds the agent may take before it must answer.
History limit `historyLimit`	`int`	`20`	How many prior turns to recall from memory.

This page is generated from the node registry by gen-node-docs.mjs on every site build — ports, properties, defaults and visibility rules cannot drift from the code.