Overview
llms.txt is an auto-generated documentation endpoint that provides structured content for AI assistants. It’s accessible at /llms.txt, /blog.md, /blog/*.md, /docs.md, and /docs/*.md.
How It Works
The endpoint automatically scans documentation and blog content, generating markdown files that include:
- Complete documentation pages
- Blog articles
- Structured navigation
- Cross-references between pages
Access Points for AI Assistants
The following endpoints are available:
/llms.txt - Index for complete documentation/blog.md - Index for latest articles and updates/llms-full.txt - Complete documentation in one page for AI assistants/docs/{page}.md - Individual documentation pages in markdown/blog/{post}.md - Individual blog posts in markdown
Implementation
The endpoint is implemented in web/next/src/app/(llms.txt)/llms.txt/[[...slug]]/route.ts:
web/next/src/app/(llms.txt)/llms.txt/[[...slug]]/route.ts
import { notFound } from "next/navigation"
import blogMeta from "@/../content/blog/meta.json"
import docsMeta from "@/../content/docs/meta.json"
import { config } from "@/lib/config"
import { sortByMeta } from "@/lib/sort-by-meta"
import { blogSource, docsSource } from "@/lib/source"
export const revalidate = false
export async function GET(_req: Request, { params }: { params: Promise<{ slug?: string[] }> }) {
const { slug } = await params
if (!slug) {
const docsPages = sortByMeta(docsSource.getPages(), docsMeta.pages, "/docs")
const docsIndex = docsPages
.map((p) => `- [${p.data.title}](${config.app.url}${p.url}.md): ${p.data.description}`)
.join("\n")
return new Response(
`# ${config.app.name}
> ${config.app.description}
## Documentation
> Complete documentation for ${config.app.name}
${docsIndex}
## Optional
- [Blog](${config.app.url}/blog.md): Latest articles and updates
`,
{
headers: {
"Content-Type": "text/markdown",
},
},
)
}
// ... rest of the implementation
}
Features
- Automatic markdown generation with links and metadata
- Uses
docsSource from Fumadocs for documentation
- Uses
blogSource from Fumadocs for blog content
- Processes markdown with
includeProcessedMarkdown: true
- Returns plain text markdown for AI consumption
Configuration
Enable processed markdown in web/next/source.config.ts:
web/next/source.config.ts
import { defineConfig, defineDocs } from "fumadocs-mdx/config"
export const docs = defineDocs({
dir: "content/docs",
docs: {
postprocess: {
includeProcessedMarkdown: true,
},
},
})
export const blog = defineDocs({
dir: "content/blog",
docs: {
postprocess: {
includeProcessedMarkdown: true,
},
},
})
export default defineConfig()
Index Response (/llms.txt)
# ZeroStarter
> Modern, type-safe SaaS starter template
## Documentation
> Complete documentation for ZeroStarter
- [Getting Started](https://example.com/docs/getting-started.md): Quick start guide
- [Architecture](https://example.com/docs/architecture.md): System architecture overview
## Optional
- [Blog](https://example.com/blog.md): Latest articles and updates
Blog Index (/blog.md)
# ZeroStarter
> Modern, type-safe SaaS starter template
## Blog
> Latest articles and updates about ZeroStarter
- [Getting Started](https://example.com/blog/getting-started.md): Introduction to ZeroStarter
- [Type-Safe APIs](https://example.com/blog/type-safe-apis.md): Building type-safe APIs
## Optional
- [Documentation](https://example.com/llms.txt): Complete documentation
Individual Page (/docs/{page}.md)
# [Page Title](https://example.com/docs/page)
Page content in markdown format...
---
> To find navigation and other pages in this documentation, fetch the llms.txt file at: https://example.com/llms.txt
Use Cases
- AI Code Assistants: Provide context about your project to AI assistants
- Documentation Scrapers: Allow automated tools to index your documentation
- LLM Training: Make your documentation accessible for language model training
- Search Engines: Provide structured content for better SEO
Best Practices
- Keep content up-to-date: The endpoint uses live data from Fumadocs sources
- Use clear descriptions: AI assistants rely on page descriptions for context
- Structure content well: Use headings and sections for better parsing
- Include metadata: Frontmatter helps AI understand page context
The llms.txt endpoint automatically updates when you add or modify documentation. No manual updates required!
Testing
Test the endpoint:
# Fetch the index
curl https://yourdomain.com/llms.txt
# Fetch blog index
curl https://yourdomain.com/blog.md
# Fetch a specific page
curl https://yourdomain.com/docs/getting-started/installation.md
SEO Benefits
- Machine-readable documentation format
- Structured content for AI assistants
- Automatic updates when content changes
- Cross-references between pages
- Clear navigation structure
Privacy
All content accessible via llms.txt endpoints is publicly available. Don’t include sensitive information in your documentation.
Resources