diff --git a/Categories.mediawiki b/Categories.mediawiki
index ff8b26d..e58b757 100644
--- a/Categories.mediawiki
+++ b/Categories.mediawiki
@@ -1,5 +1,6 @@
 <!-- DO NOT EDIT THIS FILE, CODE AUTOMATICALLY GENERATED BY category.sh -->
 The following categories contain pages or media.
+* [[Category Cache]] (1 members)
 * [[Category Data Type]] (3 members)
 * [[Category Packet]] (91 members)
 * [[Category Packet 194]] (4 members)
diff --git a/Category-Cache.mediawiki b/Category-Cache.mediawiki
index 41043c7..a905efc 100644
--- a/Category-Cache.mediawiki
+++ b/Category-Cache.mediawiki
@@ -2,4 +2,8 @@
 
 == '''JAGGRAB Protocol''' ==
 
-== '''Ondemand Protocol''' ==
\ No newline at end of file
+== '''Ondemand Protocol''' ==
+<!-- DO NOT EDIT BELOW THIS LINE, OR CHANGE THIS COMMENT, CODE AUTOMATICALLY GENERATED BY category.sh -->
+== '''Pages in category "Cache"''' ==
+The following 1 pages are in this category.
+* [[JAGGRAB Protocol]]
diff --git a/JAGGRAB-Protocol.mediawiki b/JAGGRAB-Protocol.mediawiki
new file mode 100644
index 0000000..7a244c9
--- /dev/null
+++ b/JAGGRAB-Protocol.mediawiki
@@ -0,0 +1,38 @@
+[[Category Cache]]
+
+== Introduction ==
+
+The JAGGRAB protocol is used to 'grab' cache files from the file server and download them.
+
+It is a text-based protocol, similar to HTTP, and the client will fall back to HTTP if JAGGRAB is unavailable. This generally happens in unsigned mode and helps users who are behind firewalls.
+
+== Request format ==
+
+A request is simply the text JAGGRAB, a space, the path to the file and a newline character. Therefore, it is very similar to a HTTP/0.9 GET request.
+
+ JAGGRAB /path/to/file
+
+=== New engine ===
+
+In (perhaps all) new engine clients, the client prefixes the JAGGRAB request line with a single byte (value 17).
+
+== Response format ==
+
+The response is simply the raw file data. Once the response is sent, the connection is closed.
+
+== Files ==
+
+There are a number of files which map to files in the cache.
+
+* '''/crc''' - the CRC table
+* '''/title''' - cache 0, file 1
+* '''/config''' - cache 0, file 2
+* '''/interface''' - cache 0, file 3
+* '''/media''' - cache 0, file 4
+* '''/versionlist''' - cache 0, file 5
+* '''/textures''' - cache 0, file 6
+* '''/wordenc''' - cache 0, file 7
+* '''/sounds''' - cache 0, file 8
+
+NOTE: the client will usually postfix these with random numbers, so when checking for the file only the start of the string should be examined: not the whole one.<BR>
+NOTE: The crc is postfixed with the client revision
\ No newline at end of file