Docs Lookup Agent

Overview

An AI agent workflow that answers technical questions by searching documentation using Context7 and synthesising answers with Gemini. Perfect for creating internal support bots or documentation assistants.

How It Works

1. Receive Question - Accept technical query via webhook
2. Context7 Search - Search documentation for relevant sections
3. Context Assembly - Gather relevant documentation chunks
4. Gemini Analysis - AI generates answer from context
5. Response - Return answer with documentation references

Workflow Components

Features

• Context7 Integration - Searches indexed documentation
• Conversational - Maintains context across questions
• Source Links - Points to specific documentation sections
• Fallback Handling - Graceful responses when docs don’t cover topic
• Customisable Prompt - Adjust AI personality and response format

What is Context7?

Context7 is a documentation search tool that indexes technical docs and provides semantic search. It’s particularly useful for:

• API documentation
• Framework guides
• Internal knowledge bases
• Technical wikis

Example Interaction

User: “How do I set up authentication in Next.js 14?”

Agent:

Based on the Next.js documentation, here's how to set up
authentication in version 14:

1. **Server Components** - Use `getServerSession()` from next-auth
   in Server Components for secure session access.

2. **Middleware** - Add `middleware.ts` to protect routes:
   ```ts
   export { default } from "next-auth/middleware"
   export const config = { matcher: ["/dashboard/:path*"] }

3. API Routes - Use the new App Router API route handlers at app/api/auth/[...nextauth]/route.ts

📚 Source: Next.js Authentication Guide (v14)

## Use Cases

**Internal Support** - Answer employee questions about internal tools

**Developer Onboarding** - Help new devs learn the codebase

**Customer Support** - Technical product support

**Documentation Chat** - Interactive docs experience

## Configuration

| Setting | Description |
|---------|-------------|
| Context7 Index | Which documentation to search |
| Gemini Model | gemini-pro or gemini-1.5-pro |
| System Prompt | AI personality and constraints |
| Max Context | Token limit for context window |
| Response Format | Markdown, plain text, or HTML |