llms.txt

The llms.txt file is a convention for providing machine-readable context about a website to AI assistants and large language models. Similar to how robots.txt guides search engine crawlers, llms.txt gives LLMs structured information about what a site does, what APIs it offers, and how to interact with it.

Unmarkdown™ provides two versions, each serving a different level of detail:

• Summary (llms.txt): A compact overview of Unmarkdown, its features, and API endpoints. Best for quick context when an LLM needs to know what Unmarkdown does.
• Full reference (llms-full.txt): Comprehensive documentation with all endpoint details, request/response formats, parameter tables, and usage examples. Best when generating integration code or answering detailed questions.

Both files are public and require no authentication.

# Summary version
curl https://unmarkdown.com/llms.txt

# Full version
curl https://unmarkdown.com/llms-full.txt

// JavaScript
const summary = await fetch("https://unmarkdown.com/llms.txt").then(r => r.text());
const full = await fetch("https://unmarkdown.com/llms-full.txt").then(r => r.text());

# Python
import requests

summary = requests.get("https://unmarkdown.com/llms.txt").text
full = requests.get("https://unmarkdown.com/llms-full.txt").text

Both files use Markdown with structured sections. The content is organized into clear headings so that LLMs can parse and reference specific parts efficiently. Sections include:

• Product description and positioning
• Feature overview and capabilities
• API endpoints, authentication, and rate limits
• Pricing and plan details
• Template catalog and AI action descriptions
• Links to further documentation

The llms.txt files are designed for a variety of AI-powered workflows:

• AI coding assistants: provide context about the Unmarkdown API when generating integration code
• Chatbots and agents: give AI assistants accurate, up-to-date information about Unmarkdown features
• Code generation tools: supply endpoint details and schemas for automated client library creation
• RAG pipelines: include Unmarkdown documentation as a knowledge source for retrieval-augmented generation

The files are updated alongside the API and docs. If you are fetching them programmatically, caching for up to 24 hours is reasonable. The content changes only when new features or endpoints are released.