Node-RED LLM Palette (Flowise-style Chatbot and Agent Builder)
This is an ongoing project, but I have made good progress. There are example flows and nodes in the palette. The bonus is that I learn about AI, Agents, MCP etc in an environment that I understand. The Brave and Firecrawl MCP's allow you to browse the web and get information back from within your flows. The File system MCP enables you to work with the local filesystem.
Overview
This project enables developers to build custom Chatbots and Agents inside Node-RED using OpenAI APIs.
It provides modular, composable nodes for LLM interactions — similar to FlowiseAI, but built natively for Node-RED.
Objectives
-
Build reusable Node-RED nodes for OpenAI-based LLMs.
-
Support ChatGPT-like conversational experiences.
-
Enable RAG (retrieval-augmented generation) and Agent-style tool calling.
-
Leverage Node-RED’s visual programming environment for easy orchestration.
Node Palette Overview
| Node | Function |
|---|---|
| openai-config | Stores OpenAI credentials and default settings |
| prompt-template | Templating node for dynamic prompt generation |
| openai-chat | Chat completion node (streaming and multi-turn) |
| openai-embeddings | Generates embeddings for text or documents |
| agent-orchestrator | Handles tool invocation and reasoning loops |
| memory/vectorstore | Optional persistent memory and retrieval storage |
