Seo llms txt

LLMs.txt Generator — AI Visibility for Next.js Sites

Overview

The llms.txt standard provides AI agents with structured guidance about your site. Rather than a static file that goes stale, this skill generates dynamic route handlers that pull from your actual data sources (products, services, blog posts) so the AI-facing file always reflects your current content.

How It Works

The generator creates two Next.js App Router route handlers:

Route	Purpose	Detail Level
/llms.txt	AI Agent Quick Reference	Concise — services, products, pages, contact
/llms-full.txt	Full AI Context	Comprehensive — architecture, schema, database, working instructions

Both are plain-text GET responses generated at request time from the same data sources your site already uses.

Usage

Quick Start

/seo llms-txt generate

This will:

1. Scan your project for data sources (auto-detects Next.js App Router)
2. Generate app/llms.txt/route.ts and app/llms-full.txt/route.ts
3. Verify the routes compile (runs typecheck)

Validate Existing

/seo llms-txt validate

Checks if the routes exist, are valid, and reflect current data.

Data Source Detection

The generator auto-discovers these sources:

Source	Purpose	How Detected
lib/metadata.ts	Site name, URL, owner	Grep for export const siteUrl
lib/products.ts	Product/shop listings	Grep for export const products or export type Product
lib/data/services.ts	Service definitions	Grep for export const services or export interface Service
lib/mdx.ts	Blog post fetching	Grep for getAllPosts or getPostBySlug
content/posts/	MDX blog files	Glob for *.mdx
app/sitemap.ts	Route discovery fallback	Check if sitemap exports exist

Fallback Behavior

If no matching data sources are found, the generator:

• Extracts site name from package.json or app/layout.tsx metadata
• Discovers routes from the sitemap or filesystem
• Creates a minimal but useful llms.txt with what's available

Generated Route Structure

Minimal Route (app/llms.txt/route.ts)

import { getAllPosts } from '@/lib/mdx';
import { siteUrl, siteName } from '@/lib/metadata';
import { products } from '@/lib/products';
import { services, serviceGroups } from '@/lib/data/services';

export async function GET(): Promise<Response> {
  const lines: string[] = [
    `# ${siteName}`,
    '',
    '> Description of your site...',
    '',
    '## Core Pages',
    `- [Home](${siteUrl}/)`,
    '...'
  ];

  return new Response(lines.join('\n'), {
    headers: {
      'Content-Type': 'text/plain; charset=utf-8',
      'Cache-Control': 'public, max-age=3600, s-maxage=3600'
    }
  });
}

Full Route (app/llms-full.txt/route.ts)

Extended version with:

• Business Overview — value proposition, methodology
• Full Service Details — each service with deliverables, timeline, ideal-for
• Product Tables — available + coming-soon in markdown tables
• Blog Index — all posts with dates, categories, reading times
• Technical Architecture — stack, environment variables, directory tree
• SEO & Structured Data — schema types per page, metadata system
• Database Schema — table definitions
• AI Agent Guidance — how to work on the project

Format Specification

The generated files follow the llms.txt standard:

# Site Name
> Short description

## Section
- [Link](url): Description

## Data Tables
| Col A | Col B |
|-------|-------|
| Value | Value |

Rules:

• Single # H1 per file
• ## H2 for major sections, ### H3 for subsections
• [text](url) markdown links
• Pipe tables for structured data
• Backtick code blocks for technical content
• Content-Type: text/plain; charset=utf-8

Validation

After generation, the skill verifies:

1. TypeScript — npm run typecheck (or npx tsc --noEmit)
2. Lint — npm run lint (no errors, warnings ok)
3. Route accessibility — File exists at correct path for framework
4. Content correctness — Spot-check that products match lib/products.ts, posts match MDX files

Integration with SeoFlow

This skill complements the existing seo-geo agent (which audits llms.txt presence) by providing the generation side — instead of just reporting that an llms.txt is missing, it creates one.

After generating, you can run:

• /seo geo <url> — audits the generated llms.txt for GEO compliance
• /seo audit <url> — full site audit including AI search readiness

Error Handling

Scenario	Action
Not a Next.js project	Guide user through creating a manual llms.txt (static file in public/ or root)
App Router not detected	Fall back to Pages Router API route at pages/api/llms-txt.js
No data sources	Use package.json name + sitemap routes for minimal output
Route already exists	Check if existing route is static or dynamic; offer to upgrade to dynamic if static
Typecheck fails	Report the specific error; do not commit broken routes

Example Output

For a consulting/shop hybrid site like Impact by Eri, the generated /llms.txt includes:

• Business name and tagline
• 6 core page links
• 2 service packages with 6 total services
• Products split by availability (available vs coming-soon)
• Blog posts with categories and dates
• Contact info, location, pricing range
• AI agent guidance for working on the codebase

The /llms-full.txt expands to include full product tables (12+ products with record counts, prices, sectors), technical stack (Next.js 15, Supabase, Stripe, etc.), directory structure, and database schema.