Even if you have no platform points left, you can connect your own LLM APIs. WorkBuddy supports integrating external LLM APIs, Coding Plans, and Token Plans.
Enter from settings, select model options, click "Add Model", and choose from various coding plans or custom APIs.
For instance, for DeepSeek, you only need to enter your API key. Or, for a local Ollama model, start Ollama locally (default port 11434, OpenAI-compatible endpoint). Local models ensure that data never leaves your computer, support offline usage, and cost zero tokens.
Core Configuration config.json Explained
WorkBuddy registers Node.js-based MCP connectors by reading its core global configuration file config.json. This file is typically located at C:\\Users\\Username\\.workbuddy\\config.json (Windows).
Notion MCP Server Configuration Example
- Go to the Notion Developer console, create an internal integration, and get the
Integration Token. - Open
config.jsonand register the official@notionhq/notion-mcp-serverunder themcpServersnode:
{
"mcpServers": {
"notion": {
"command": "npx",
"args": [
"-y",
"@notionhq/notion-mcp-server"
],
"env": {
"NOTION_API_TOKEN": "your_notion_token_here"
}
}
}
}- Open the Notion page you want to authorize, click the "..." icon in the top right corner, select "Connect to", and choose the integration name you created.
- Restart the WorkBuddy client to reload the configuration file and apply changes.
[!CAUTION] Context Window Overflow warning: Each registered MCP server exposes several or dozens of tools descriptions. Loading too many MCP servers causes the system prompt context to expand rapidly, increasing API costs and potentially degrading task planning accuracy due to tool noise. Enable only the MCP servers you currently need, and disable unused ones.
References
Part of this series
WorkBuddy In Action
View all guides in this series →