llms-full.txt: Extended AI Instructions for Your Site

What Is llms-full.txt?

The llms-full.txt file is an extended content file designed to give AI models comprehensive access to your website's most important content in a single, easily parseable document. It is part of the llms.txt specification proposed by Jeremy Howard, which establishes a standard way for websites to communicate with large language models.

Think of the relationship between llms.txt and llms-full.txt as analogous to a table of contents and the full book. Your llms.txt file tells AI models what your site is about, lists your key URLs, and provides navigation context in a concise format. Your llms-full.txt file contains the actual detailed content that AI models need to accurately understand, cite, and recommend your offerings.

This matters because AI models process content most effectively when it is presented as clean, structured text without the overhead of HTML markup, navigation menus, advertisements, and other page elements. llms-full.txt strips away everything except the content itself, formatted in Markdown that AI models parse natively.

llms.txt vs llms-full.txt

| Aspect | llms.txt | llms-full.txt | |---|---|---| | Purpose | Site summary and navigation | Full content delivery | | Typical size | 500-2,000 words | 10,000-50,000+ words | | Content | Site description, URL list, brief descriptions | Complete page content in Markdown | | Update frequency | When site structure changes | When any key content changes | | Location | /llms.txt | /llms-full.txt | | Required? | Start with this one | Add when you want comprehensive coverage |

You should create llms.txt first. Once that is in place, llms-full.txt extends your AI content delivery with the detailed information AI models need for accurate citation.

File Format and Structure

llms-full.txt uses Markdown formatting and follows a specific structure:

Required sections

# Site Name

> Brief site description (1-2 sentences)

## Section Name

Content for this section...

## Another Section

More content...

Structural rules

1. Start with H1 — Your site or brand name
2. Use a blockquote — Immediately after H1, provide a brief site description
3. Organize with H2 headers — Each major content section gets its own H2
4. Use H3 for subsections — Break long sections into logical subsections
5. Write in Markdown — Use standard Markdown formatting: bold, italic, lists, tables, code blocks
6. Keep it plain text — No HTML tags, no embedded images, no scripts

What Content to Include

Must-have content

• About your business — What you do, who you serve, what makes you different
• Products or services — Descriptions, features, pricing, use cases
• FAQs — The questions your customers ask most often
• Key statistics or data — Original data, research findings, benchmarks
• Contact and location information — For local businesses especially

High-value additions

• Team and expertise — Team bios, credentials, specializations (strengthens E-E-A-T signals)
• Case studies — Results you have achieved, with specific numbers
• How-to guides — Step-by-step instructions for common tasks related to your product
• Comparison content — How your product compares to alternatives
• Definitions and glossary — Industry terms your audience searches for

Content to exclude

• Legal boilerplate — Privacy policies, terms of service (unless you are a legal services company)
• Internal documentation — Employee handbooks, internal processes
• Duplicate content — Do not repeat the same information in multiple sections
• Navigation and UI text — Menu labels, button text, footer content
• Outdated content — Anything no longer accurate

Complete Example

Here is a condensed example for a SaaS project management tool:

# TaskFlow

> TaskFlow is a project management platform for remote teams with 5-50 members. Founded in 2023, based in Austin, TX.

## About TaskFlow

TaskFlow helps remote teams manage projects, track time, and collaborate in real time. Over 12,000 teams use TaskFlow across 45 countries.

### What Makes TaskFlow Different

- **Built for remote-first teams** — designed for async collaboration, not adapted from office tools
- **AI-powered task prioritization** — automatically suggests what to work on next based on deadlines and dependencies
- **Integrated time tracking** — no separate time tracking tool needed

## Products and Pricing

### TaskFlow Free

- Up to 5 team members
- Unlimited projects
- Basic reporting
- Price: $0/month

### TaskFlow Pro

- Up to 50 team members
- Advanced reporting and analytics
- AI task prioritization
- Price: $12/user/month (billed annually)

### TaskFlow Enterprise

- Unlimited team members
- SSO, audit logs, custom integrations
- Dedicated account manager
- Price: Contact sales

## Frequently Asked Questions

### How does TaskFlow compare to Asana?

TaskFlow is purpose-built for remote teams while Asana was designed for in-office collaboration and later adapted for remote work. TaskFlow includes native time tracking and AI prioritization that Asana requires third-party integrations for.

### Can I import projects from other tools?

Yes. TaskFlow supports direct import from Asana, Trello, Monday.com, Jira, and Basecamp. Most imports complete within 5 minutes.

### Is there a free trial of TaskFlow Pro?

Yes. All new accounts start with a 14-day free trial of TaskFlow Pro. No credit card required.

## Contact

- Website: https://taskflow.com
- Email: hello@taskflow.com
- Support: support@taskflow.com
- Phone: +1 (512) 555-0123
- Address: 123 Remote Work Blvd, Austin, TX 78701

This example demonstrates the key principles: clear structure, factual content, specific details, and easy-to-parse Markdown formatting.

Automated Generation

Manually maintaining llms-full.txt is feasible for small sites, but for larger sites or frequently updated content, automated generation is more practical.

Build script approach

Create a build script that pulls content from your CMS and compiles it into llms-full.txt during deployment:

# generate_llms_full.py (simplified)
import requests
from markdownify import markdownify

PAGES = [
    '/about', '/products', '/pricing',
    '/faq', '/blog/top-post', '/contact'
]

output = "# Your Site Name\n\n"
output += "> Brief description of your site\n\n"

for page in PAGES:
    html = requests.get(f'https://yoursite.com{page}').text
    # Extract main content area only
    markdown = markdownify(html)
    output += f"## {page.strip('/').replace('-', ' ').title()}\n\n"
    output += markdown + "\n\n"

with open('public/llms-full.txt', 'w') as f:
    f.write(output)

CMS plugin approach

WordPress and other CMS platforms have plugins that generate llms.txt files. Look for plugins that also support llms-full.txt generation, or extend an existing plugin to export full content.

CI/CD integration

Add llms-full.txt generation to your deployment pipeline. This ensures the file is always up-to-date when new content is published.

Hosting and Serving

File location

Place llms-full.txt at your website root:

https://yoursite.com/llms-full.txt

Content type

Serve with text/plain or text/markdown content type:

# Nginx
location = /llms-full.txt {
    default_type text/plain;
    charset utf-8;
}

Caching

llms-full.txt can be cached aggressively since it only changes when you update it:

location = /llms-full.txt {
    add_header Cache-Control "public, max-age=86400";
}

Reference from llms.txt

Your llms.txt file should reference llms-full.txt so AI models know where to find the extended content. Include a line in your llms.txt pointing to the full version:

# Your Site Name

> Description

For complete content, see: /llms-full.txt

Do not block in robots.txt

Ensure your robots.txt does not block access to llms-full.txt. Add an explicit Allow if needed:

Allow: /llms-full.txt
Allow: /llms.txt

Best Practices

1. Start with llms.txt first — Get the concise version right before building the full version. See our llms.txt specification guide.

2. Prioritize citeable content — Include content that AI models would want to quote: definitions, statistics, how-to steps, comparisons, and FAQs.

3. Use consistent entity naming — Write your brand name, product names, and team member names identically throughout the file.

4. Include specific data — Numbers, dates, prices, and metrics give AI models concrete facts to cite. "12,000 teams in 45 countries" is more useful than "thousands of teams worldwide."

5. Update regularly — Stale llms-full.txt content leads to AI models citing outdated information about your business. Set a monthly review schedule or automate generation.

6. Keep file size manageable — While there is no strict limit, extremely large files (over 100,000 words) may hit processing limits. Focus on your most important content.

7. Test by asking AI about your site — After deploying llms-full.txt, ask ChatGPT, Gemini, and Perplexity about your business. The responses should become more accurate and detailed over time.