From e7c6f128b32abfa2bfcf7a18579af3ebcaa2e61e Mon Sep 17 00:00:00 2001
From: Logan Besecker <me@LoganBesecker.com>
Date: Thu, 1 Oct 2026 09:45:48 -0700
Subject: [PATCH] Treat a JSON page at /llms.txt as absent

Of 1,988 llms.txt files fetched in production, 1,962 open with a Markdown
heading as the format specifies. Four are JSON: API hosts answering every
path with a status document, served as text/plain.

The HTML check caught single-page-app fallbacks but not these. Neither
llms.txt nor AGENTS.md is JSON, and a status page carrying a timestamp
would record a fake change on every weekly check.

- A JSON content type, or a body that parses as JSON, counts as absent
- Test reproduces the production case

Closes #81

---

Pages affected:

- [MCP server changelog](https://ai.mcpharbor.dev/changelog) -- kept free of fake llms.txt changes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 lib/mcp_registry/documents.ex        | 12 +++++++++++-
 test/mcp_registry/documents_test.exs |  7 +++++++
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/lib/mcp_registry/documents.ex b/lib/mcp_registry/documents.ex
index c15c22f..b825805 100644
--- a/lib/mcp_registry/documents.ex
+++ b/lib/mcp_registry/documents.ex
@@ -414,7 +414,17 @@ defmodule McpRegistry.Documents do
       |> String.downcase()
 
     String.valid?(body) and not String.contains?(type, "html") and
-      not String.starts_with?(head, ["<!doctype", "<html", "<?xml", "<head"])
+      not String.starts_with?(head, ["<!doctype", "<html", "<?xml", "<head"]) and
+      not json?(type, body)
+  end
+
+  # The other catch-all: an API host answering every path with a JSON status
+  # document. Neither llms.txt nor AGENTS.md is JSON, and one that carries a
+  # timestamp would otherwise "change" on every check.
+  defp json?(type, body) do
+    String.contains?(type, "json") or
+      (String.starts_with?(String.trim_leading(body), ["{", "["]) and
+         match?({:ok, _}, Jason.decode(body)))
   end
 
   defp header(%Req.Response{} = resp, name) do
diff --git a/test/mcp_registry/documents_test.exs b/test/mcp_registry/documents_test.exs
index b0430ec..3d07f70 100644
--- a/test/mcp_registry/documents_test.exs
+++ b/test/mcp_registry/documents_test.exs
@@ -142,6 +142,13 @@ defmodule McpRegistry.DocumentsTest do
       assert [%{fields: %{"file" => ["absent", "published"]}}] = changes()
     end
 
+    test "a JSON status page served at /llms.txt is not an llms.txt" do
+      # Seen in production: an API host answering every path with its health
+      # document, as text/plain.
+      serve(&text(&1, 200, ~s({"status":"ok","server":"mock-mcp-server"})))
+      assert Documents.check(doc()) == :missing
+    end
+
     test "an outage is not a removal" do
       d = doc()
       serve(&text(&1, 200, "# Acme\n"))