If you dont give a shit about agent, u can throw msg.payload to this node and it will send it to deepseek.

Node-RED factory-agent-deepseek

A Node-RED node for interacting with Deepseek's AI API to create agent-based systems. This node is designed to simplify the process of sending prompts to Deepseek's models and handling responses within Node-RED flows.

Features

• Compose and send requests to Deepseek API with minimal configuration
• Support for both deepseek-chat and deepseek-reasoner models
• Combines environment, state, and action data into JSON format for user messages
• Configurable temperature and max tokens settings
• Secure API key storage

Installation

Install via Node-RED's Manage Palette:

1. In Node-RED, navigate to the menu → Manage palette
2. Go to the "Install" tab
3. Search for "factory-agent-deepseek"
4. Click install

Or install via npm:

cd ~/.node-red
npm install factory-agent-deepseek

Usage

Node Configuration

1. Drag the factory-agent-deepseek node from the factory category in the palette to your flow
2. Double-click the node to open the configuration panel
3. Enter your Deepseek API key directly in the node configuration
4. Select the model (deepseek-reasoner or deepseek-chat)
5. Adjust temperature and max tokens settings if needed
6. Click Done to save

Input

The node expects the following input properties:

• msg.sysPrompt (string): The system prompt to send to Deepseek's system role
• msg.envPrompt (string): Environment prompt to be combined with state and action
• msg.state (string): State information to be combined with envPrompt and action
• global.action (string): Action information from global context

Output

The node outputs:

• msg.payload (object): The complete response from Deepseek API
• msg.deepseekRequest (object): The request that was sent to Deepseek API
• msg.result (string): The content from the Deepseek response message
• msg.reasoning_content (string): The reasoning_content from the Deepseek response (if available)

Flow Context

The node manages a flow context variable:

• flow.agentstate: Tracks the request status
  • processing: Set when the request is sent to Deepseek API
  • received: Set when a successful response is received

Example Flow

[{"id":"f6f2187d.f17ca8","type":"factory-agent-deepseek","z":"34de40b2.eb705","name":"Deepseek Agent","model":"deepseek-reasoner","temperature":"0.7","maxTokens":"2048","x":450,"y":300,"wires":[["c234a454.0cb648"]],"credentials":{"apiKey":"YOUR_API_KEY_HERE"}},{"id":"5a35c1ae.5e231","type":"inject","z":"34de40b2.eb705","name":"Test Input","props":[{"p":"sysPrompt","v":"You are a helpful assistant.","vt":"str"},{"p":"envPrompt","v":"Current time: 14:30","vt":"str"},{"p":"state","v":"User asked about weather.","vt":"str"}],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","x":270,"y":300,"wires":[["f6f2187d.f17ca8"]]},{"id":"c234a454.0cb648","type":"debug","z":"34de40b2.eb705","name":"Response","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"payload","targetType":"msg","statusVal":"","statusType":"auto","x":640,"y":300,"wires":[]}]

Files

The node includes the following files:

1. factory-agent-deepseek.js - Main implementation
2. factory-agent-deepseek.html - Configuration UI and help documentation
3. package.json - Node-RED package information
4. README.md - Documentation

License

MIT