The site-wide files

Three files, three lines each.

app/llms.txt/route.ts

import { createGeoffyTextRoute } from "@geoffy/headless/next";
export const GET = createGeoffyTextRoute({ siteKey: process.env.GEOFFY_SITE_KEY! }, "llms.txt");

app/llms-full.txt/route.ts

import { createGeoffyTextRoute } from "@geoffy/headless/next";
export const GET = createGeoffyTextRoute({ siteKey: process.env.GEOFFY_SITE_KEY! }, "llms-full.txt");

app/agents.md/route.ts

import { createGeoffyTextRoute } from "@geoffy/headless/next";
export const GET = createGeoffyTextRoute({ siteKey: process.env.GEOFFY_SITE_KEY! }, "agents.md");

All three serve the same document

One body, three names. Three conventions exist for “the file an AI agent should read first”, so we answer to all three rather than making you pick — but there is one document, not three. Nothing is lost by mounting all of them, and mounting only one means agents that look for the other names find nothing.

Why route handlers and not a rewrite

A rewrite in next.config.js works, but it must be in the beforeFiles array.

If your app has a catch-all segment such as app/[lang], a rewrite registered anywhere else loses to it — and the catch-all answers /llms.txt with your home page and HTTP 200, which a crawler reads as a successful answer rather than a missing file.

Route handlers cannot lose that race. app/llms.txt/route.ts is a static route segment, and static segments beat dynamic ones with no configuration.

This failure looks like success

A 404 would be obvious. A 200 carrying your home page is not: your llms.txt “works”, and what the crawler reads is your marketing copy. If you already have a rewrite, delete it when you add these files.

Tuning the cache

The fetch is cached for revalidateSeconds, one hour by default, under the helper’s own cache tag:

createGeoffyTextRoute({ siteKey, revalidateSeconds: 300 }, "llms.txt");

Mounting the revalidate route purges that tag on publish — see how updates arrive.

If Geoffy is unreachable

The route answers 503 with a Retry-After header, never a fabricated 404. A crawler retries a 503; a 404 tells it to drop the page from its index, which is a much more expensive mistake to make on your behalf.