Goldnat AI Connect for XenForo Pro [Paid] 1.2.14

[chanina]

canina submitted a new resource:

AI Connect Pro for XenForo - WebMCP Bridge - Expose your XenForo forum as structured MCP tools for AI agents

AI Connect Pro extends the free AI Connect for XenForo plugin with additional write capabilities, allowing AI agents (Claude, ChatGPT, Copilot, Gemini) to interact with your forum via structured MCP tools over HTTP.


Pro capabilities:

• Browse and list forums
• Create new threads
• Reply to existing threads
• Edit posts
• Send private conversations
• Priority support...

Read more about this resource...

 

[chanina]

canina updated Goldnat AI Connect for XenForo Pro with a new update entry:

From 5 tools to 67, with per-tool permissions

From 5 tools to 67, with per-tool permissions​

Version 1.0.5 shipped five write tools. This release has 67, organised into eight groups, each with its own permission entry.

If you already own this add-on, the upgrade is free. Details at the end of this post.

What the add-on can now do​

Group	Tools	Covers
Moderation	14	Lock, stick, feature, move, delete, mark solutions...

Read the rest of this update entry...

 

[IMCD]

I posted this in the non-pro resource as well:

Suggestion:

Something like conversation:read access and a listConversations in order to pull recent conversations and interact with those. Currently, no way to pull a list, only interact with known conversation IDs.

Also, any eta on the moderation version? Would be a huge help.

 

[IMCD]

Suspected bug found:

Server error: TokenRegistry cleanup calls a method XenForo doesn't have

On API requests (for example GET /forums/api/aiconnect-tools), after a valid bearer token is checked, OAuthServer::validateToken() lazily runs TokenRegistry::runCleanup() if cleanup hasn’t run in 24 hours. That throws:

Error: TokenRegistry::runCleanup failed: Call to undefined method XF\Db\Mysqli\Adapter::affectedRows()
src/addons/chgold/AIConnect/Repository/TokenRegistry.php:402
XF\Db\Mysqli\Adapter has no affectedRows(). $db->query() returns a statement. The count is $stmt->rowsAffected(). XenForo core does it that way (AbstractAdapter::update(), FloodCheck).

The same bad call is also in revokeUnused() and revokeInactive().

Because the exception is caught before last_cleanup_at is written, cleanup never finishes and the error repeats on later API calls.

Fix: keep the statement and use rowsAffected():

$stmt = $db->query('UPDATE ...');
$affected = (int)$stmt->rowsAffected();
File: src/addons/chgold/AIConnect/Repository/TokenRegistry.php
Methods: runCleanup(), revokeUnused(), revokeInactive() (six call sites).

 

[chanina]

I posted this in the non-pro resource as well:

Suggestion:

Something like conversation:read access and a listConversations in order to pull recent conversations and interact with those. Currently, no way to pull a list, only interact with known conversation IDs.

Also, any eta on the moderation version? Would be a huge help.

Suspected bug found:

Server error: TokenRegistry cleanup calls a method XenForo doesn't have

On API requests (for example GET /forums/api/aiconnect-tools), after a valid bearer token is checked, OAuthServer::validateToken() lazily runs TokenRegistry::runCleanup() if cleanup hasn’t run in 24 hours. That throws:

Error: TokenRegistry::runCleanup failed: Call to undefined method XF\Db\Mysqli\Adapter::affectedRows()
src/addons/chgold/AIConnect/Repository/TokenRegistry.php:402
XF\Db\Mysqli\Adapter has no affectedRows(). $db->query() returns a statement. The count is $stmt->rowsAffected(). XenForo core does it that way (AbstractAdapter::update(), FloodCheck).

The same bad call is also in revokeUnused() and revokeInactive().

Because the exception is caught before last_cleanup_at is written, cleanup never finishes and the error repeats on later API calls.

Fix: keep the statement and use rowsAffected():

$stmt = $db->query('UPDATE ...');
$affected = (int)$stmt->rowsAffected();
File: src/addons/chgold/AIConnect/Repository/TokenRegistry.php
Methods: runCleanup(), revokeUnused(), revokeInactive() (six call sites).

Thanks for the kind words — really appreciated!

listConversations — This should be fixed in the next update.

Also: good catch on the TokenRegistry bug — affectedRows() on the adapter instead of rowsAffected() on the statement. Fixed in runCleanup(), revokeUnused(), and revokeInactive(). Will ship in the next release.

On the moderation/admin pack: it's in active development. Rather than a single tool, we're shipping a full Users & Groups bundle: list members in a group, list usergroups, get a single group, and a general user search. The members tool returns raw data (user_id, username, post_count, register_date, last_activity, group memberships) and lets the AI do the analysis — exactly the audit workflow you described. Landing in Pro. No date yet, but soon. I'll post here when it's live.

 

[chanina]

chanina updated Goldnat AI Connect for XenForo Pro with a new update entry:

Security hardening + audit-log analytics — 1.4.28 to 1.4.35

Eight versions, mostly security hardening on fields that agents can write to.
The headline feature is a new 4-tool bundle for analysing the AI Connect audit log itself.

If you already own this add-on, the upgrade is free.


Security — defense-in-depth on agent-writable config

v1.4.35 (Oct 26) — censorWords: extended silent-drop detection to partial drops.
If an agent submits 5 words and only 3 get stored (because XenForo rejected 2),
the tool now returns validation_failed with a count...

Read the rest of this update entry...

 

[chanina]

chanina updated Goldnat AI Connect for XenForo Pro with a new update entry:

From 67 to 84 tools, DMs added, security hardening — 1.0.40 to 1.2.14

If you already own this add-on, the upgrade is free.


What's new

17 new tools, bringing the total from 67 to 84, organised across 14 permission groups.

The new tools cover:
• Users & Groups — user lookup and group listing
• Watched & Bookmarks — follow state and bookmark lists
• Activity & Analytics — board-level engagement stats
• Forum Watch — watch/unwatch nodes

Direct Messages (Conversations) — 9 tools for reading, replying, inviting, dismissing, and creating DMs. These...

Read the rest of this update entry...

 

[CrazyChef]

I've been going through the latest AI Connect Pro 1.2.14 package before upgrading from 1.2.11. It's good to see that the previously reported fixes for attachment metadata, abstract filesystem storage, and unread conversation status appear to be included.

There are a few things I'd like to clarify before installing the update.

First, I noticed that five components present in 1.2.11 are missing from 1.2.14: Moderation, Warnings, Reports, Approval Queue, and Advanced Moderation. Was this intentional? Have these capabilities been moved into another add-on, replaced, or discontinued? The current product description lists 84 tools across 14 bundles, but the new package metadata still describes 81 tools across 11 bundles. Which reflects the actual current feature set?

Second, what is the recommended Free/Core version to use with Pro 1.2.14? Is 1.2.57 still fully supported, or should we upgrade to a newer Core release first? I'd also like to confirm that upgrading preserves existing permissions, OAuth tokens, licensing, and per-tool restrictions. In particular, do newly introduced tools remain disabled until explicitly authorized?

I'm also interested in how the add-on handles multiple AI accounts with different responsibilities. Can each account be restricted independently to specific tools and its existing XenForo permissions, with no possibility of accessing another member's private conversations or performing administrative actions without the appropriate authorization?

Another question concerns additional XenForo content types. Does Pro currently provide any support for XenForo Resource Manager, XenForo Media Gallery, or third-party add-ons, or are its tools specifically limited to native forum, conversation, profile, and attachment functionality? If those integrations aren't currently supported, are there plans to expand that functionality?

Finally, regarding security, does URL-based attachment uploading include protections against accessing internal network addresses, unsafe redirects, oversized downloads, and requests that take excessively long to complete?

I also noticed that icon.png doesn't match its entry in hashes.json, although the same discrepancy exists in the earlier package. Is that a known packaging issue?

I'm asking because I'd like to understand the supported functionality and upgrade requirements before making changes to an existing installation.

Thanks!