AI Assistant Help (RAGbot)
Learn best practices for using RAGbot—our embedded site assistant—and connecting our knowledge base to your local developer environment via Model Context Protocol (MCP).
💬 Using the RAGbot Chat Widget
The floating assistant in the bottom right corner allows you to query site documentation directly in real time.
Best Practices for Querying
- Aim for Specificity: Phrase questions with concrete technical nouns (e.g., “How do I configure the Gemini Quality Guardrails PR trigger?” rather than “How do guardrails work?”).
- Include Error Codes: If you are debugging, include specific status codes (e.g.,
429,503) or error messages. - Direct Language: Ask clear, active questions without conversational fluff for optimal vector search matching.
🔌 Connecting via Model Context Protocol (MCP)
You can connect your local AI workspace (Cursor, Claude Desktop, VS Code) directly to our documentation backend using our MCP bridge.
Quick Start with Cursor
- Go to Cursor Settings ➔ Features ➔ MCP Servers.
- Click + Add New MCP Server.
- Set Type to
command. - Enter command:
node /path/to/your/mcp-bridge/my-mcp-bridge.mjs.
MCP Server Details
- Server URL:
https://mcp.thetechnicalwriter.com - Exposed Tool:
search_documentation - Input Schema:
{ "query": "string" }