Paste-into-Chat — copy this app's docs into an LLM

One-shot HTTP fetch into an LLM context window. /llms.txt and

/<page>/llms.txt return the page's LLMS_DOC verbatim.

What this audience needs

A human sitting in front of Claude.ai, ChatGPT, or an in-product assistant wants to tell the model what your app is about. They need the prose in a known location, in a paste-friendly format, with the smallest possible friction between "I want context" and "the model has it."

For this audience, 2.0 serves:

1. /llms.txt — the site-wide prose (the home page's LLMS_DOC).
2. /<page>/llms.txt — every page's prose at a predictable URL.
3. A <link rel="alternate" type="text/markdown" href="/llms.txt">

in every HTML response so AI-aware browsers can find it.

How to use it

Direct paste. Open /llms.txt (or any per-page variant) in your browser, select all, paste into the chat.

Direct fetch. Most chat clients accept URLs as context — paste the URL itself and let the assistant fetch it.

Headless / programmatic. Pipe the content from curl:

curl https://myapp.com/llms.txt | pbcopy           # macOS
curl https://myapp.com/llms.txt | xclip            # Linux

# Or stuff it straight into a chat client that accepts stdin:
curl https://myapp.com/audiences/llm-context/llms.txt | \
    claude "Explain how this audience differs from MCP clients"

Why prose, not structured data

This audience does NOT want JSON, TOON, or MCP JSON-RPC. They want flat markdown that the model will treat as natural-language context. Structured formats are for programmatic consumers (MCP clients, crawlers parsing sitemaps). Humans paste prose.

The copy buttons on this page

The buttons next to each page entry copy that page's LLMS_DOC string directly to your clipboard, byte-for-byte identical to what you'd get from curl /<page>/llms.txt.