Know every vulnerabilitybefore it knows you.
DevGuard continuously monitors your dependencies and alerts you when CVEs like this one affect your stack — with real-time threat intelligence built for developers.
GHSA-73x5-h92w-xc2j
No affected components available
Summary
A normal authenticated user can read the content of a message in a private channel they do not belong to. GET /api/v1/channels/{id}/messages/{message_id}/thread authorizes the caller against the URL channel, but the underlying thread lookup loads the thread parent by id and returns it without verifying the parent belongs to that channel. By requesting a thread in a channel they can access while supplying a victim channel's message id as the thread root, the attacker receives the victim message — content, channel id, and author.
Affected component
backend/open_webui/models/messages.py—get_messages_by_parent_id()backend/open_webui/routers/channels.py—get_channel_thread_messages()(read),new_message_handler()(parent/reply binding on write)
Root cause
get_messages_by_parent_id(channel_id, parent_id) filters the thread replies by channel_id, but loads the thread parent by id alone and appends it without requiring parent.channel_id == channel_id:
message = await db.get(Message, parent_id) # loaded by id only — no channel binding
if not message:
return []
# replies are filtered by channel_id ...
if len(all_messages) < limit:
all_messages.append(message) # parent appended unconditionally
get_channel_thread_messages() authorizes only the URL channel, then calls get_messages_by_parent_id(id, message_id) with the caller-supplied message_id. The reply insert path (new_message_handler → insert_new_message) also stored a caller-supplied parent_id without binding it to the channel.
Impact
A non-member can disclose the content (plus channel id and author metadata) of a private-channel message whose id they know or obtain. Direct reads of the victim channel/message/thread return 403; the disclosure is via the thread parent of a channel the attacker can access. Read-only, one message per known id.
Proof of Concept
(reporter) Validated on v0.9.6: GET /channels/{attacker_channel}/messages/{victim_message_id}/thread returned the victim's private message — content, victim channel id, and author — although direct reads of the victim channel returned 403.
Fix
Bind the thread parent to the requested channel: get_messages_by_parent_id returns [] unless the parent exists and parent.channel_id == channel_id. Defence-in-depth on the write path: new_message_handler rejects a supplied parent_id/reply_to_id whose message does not belong to the URL channel.
Affected / Patched
- Affected:
< 0.10.0(last affected release 0.9.6) - Patched: v0.10.0 (PR #25766).
get_messages_by_parent_idbinds the thread parent to the requested channel (returns[]unlessparent.channel_id == channel_id), andnew_message_handlerrejects a caller-suppliedparent_id/reply_to_idwhose message does not belong to the channel.
The vulnerability can be exploited over the network without needing physical access. It is difficult for an attacker to exploit this vulnerability and may require special conditions. An attacker needs basic access or low-level privileges. No user interaction is needed for the attacker to exploit this vulnerability. The impact is confined to the system where the vulnerability exists. There is a low impact on the confidentiality of the information.
Limited exploitation activity has been observed. Close monitoring and planned remediation are recommended.
The exploit probability is very low. The vulnerability is unlikely to be exploited in the next 30 days.
We did not find any exploit available. Neither in GitHub repositories nor in the Exploit-Database.
Browse More
Continuously monitor your dependencies and get alerted when vulnerabilities like this one affect your stack.
Checkout DevGuard