discourse-ai admin-initiated SSRF when interacting with AI services

discourse-ai is the AI plugin for the open-source discussion platform Discourse. Prior to commit 94ba0dadc2cf38e8f81c3936974c167219878edd, interactions with different AI services are vulnerable to admin-initiated SSRF attacks. Versions of the plugin that include commit 94ba0dadc2cf38e8f81c3936974c167219878edd contain a patch. As a workaround, one may disable the discourse-ai plugin.

Discourse BBCode plugin vulnerable to arbitrary CSS injection

discourse-bbcode is the official BBCode plugin for Discourse. Prior to commit 91478f5, CSS injection can occur when rendering content generated with the discourse-bccode plugin. This vulnerability only affects sites which have the discourse-bbcode plugin installed and enabled. This issue is patched in commit 91478f5. As a workaround, ensure that the Content Security Policy is enabled and monitor any posts that contain bbcode.

Disclosure of the existence of secret subcategories in Discourse

Discourse is an open source platform for community discussion. In affected versions an attacker can learn that a secret subcategory exists under a public category which has no public subcategories. The issue is patched in the latest stable, beta and tests-passed version of Discourse. Users are advised to upgrade. There are no known workarounds for this vulnerability.

Arbitrary HTML injection in discourse-mermaid-theme-component

Discourse Mermaid (discourse-mermaid-theme-component) allows users of Discourse, open-source forum software, to create graphs using the Mermaid syntax. Users of discourse-mermaid-theme-component version 1.0.0 who can create posts are able to inject arbitrary HTML on that post. The issue has been fixed on the `main` branch of the GitHub repository, with 1.1.0 named as a patched version. Admins can update the theme component through the admin UI. As a workaround, admins can temporarily disable discourse-mermaid-theme-component.

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 2.9.0.beta16 on the `beta` and `tests-passed` branches, parsing posts can be susceptible to regular expression denial of service (ReDoS) attacks. This issue is patched in versions 2.8.14 and 2.9.0.beta16. There are no known workarounds.

Discourse vulnerable to bypass of post max_length using HTML comments

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 2.9.0.beta16 on the `beta` and `tests-passed` branches, users can create posts with raw body longer than the `max_length` site setting by including html comments that are not counted toward the character limit. This issue is patched in versions 2.8.14 and 2.9.0.beta16. There are no known workarounds.

Group SMTP user emails are exposed in CC email header

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 2.9.0.beta15 on the `beta` and `tests-passed` branches, recipients of a group SMTP email could see the email addresses of all other users inside the group SMTP topic. Most of the time this is not an issue as they are likely already familiar with one another's email addresses. This issue is patched in versions 2.8.14 and 2.9.0.beta15. The fix is that someone sending emails out via group SMTP to non-staged users masks those emails with blind carbon copy (BCC). Staged users are ones that have likely only interacted with the group via email, and will likely include other people who were CC'd on the original email to the group. As a workaround, disable group SMTP for any groups that have it enabled.

Discourse vulnerable to private topic leak via email#send_digest

In version 2.9.0.beta14 of Discourse, an open-source discussion platform, maliciously embedded urls can leak an admin's digest of recent topics, possibly exposing private information. A patch is available for version 2.9.0.beta15. There are no known workarounds for this issue.

Discourse password reset link can lead to in account takeover if user changes to a new email

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 3.0.0.beta16 on the `beta` and `tests-passed` branches, when a user requests for a password reset link email, then changes their primary email, the old reset email is still valid. When the old reset email is used to reset the password, the Discourse account's primary email would be re-linked to the old email. If the old email address is compromised or has transferred ownership, this leads to an account takeover. This is however mitigated by the SiteSetting `email_token_valid_hours` which is currently 48 hours. Users should upgrade to versions 2.8.14 or 3.0.0.beta15 to receive a patch. As a workaround, lower `email_token_valid_hours ` as needed.

Discourse vulnerable to exposure of user post counts per topic to unauthorized users

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 3.0.0.beta16 on the `beta` and `tests-passed` branches, the number of times a user posted in an arbitrary topic is exposed to unauthorized users through the `/u/username.json` endpoint. The issue is patched in version 2.8.14 and 3.0.0.beta16. There is no known workaround.

Discourse vulnerable to Cross-site Scripting through pending post titles descriptions

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 3.0.0.beta16 on the `beta` and `tests-passed` branches, pending post titles can be used for cross-site scripting attacks. Pending posts can be created by unprivileged users when a category has the "require moderator approval of all new topics" setting set. This vulnerability can lead to a full XSS on sites which have modified or disabled Discourse’s default Content Security Policy. A patch is available in versions 2.8.14 and 3.0.0.beta16.

Discourse vulnerable to Cross-site Scripting through tag descriptions

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 3.0.0.beta16 on the `beta` and `tests-passed` branches, tag descriptions, which can be updated by moderators, can be used for cross-site scripting attacks. This vulnerability can lead to a full XSS on sites which have modified or disabled Discourse’s default Content Security Policy. Versions 2.8.14 and 3.0.0.beta16 contain a patch.

Discourse vulnerable to Cross-site Scripting in local oneboxes

Discourse is an open source platform for community discussion. Versions prior to 2.8.13 (stable), 3.0.0.beta16 (beta) and 3.0.0beta16 (tests-passed), are vulnerable to cross-site Scripting. A maliciously crafted URL can be included in a post to carry out cross-site scripting attacks on sites with disabled or overly permissive CSP (Content Security Policy). Discourse's default CSP prevents this vulnerability. This vulnerability is patched in versions 2.8.13 (stable), 3.0.0.beta16 (beta) and 3.0.0beta16 (tests-passed). As a workaround, enable and/or restore your site's CSP to the default one provided with Discourse.

Discourse subject to Allocation of Resources Without Limits or Throttling

Discourse is an open source platform for community discussion. Versions prior to 3.0.1 (stable), 3.1.0.beta2 (beta), and 3.1.0.beta2 (tests-passed) are subject to Allocation of Resources Without Limits or Throttling. As there is no limit on data contained in a draft, a malicious user can create an arbitrarily large draft, forcing the instance to a crawl. This issue is patched in versions 3.0.1 (stable), 3.1.0.beta2 (beta), and 3.1.0.beta2 (tests-passed). There are no workarounds.

Discourse membership requests lack character limit

Discourse is an open-source discussion platform. Prior to version 3.0.1 on the `stable` branch and 3.1.0.beta2 on the `beta` and `tests-passed` branches, when submitting a membership request, there is no character limit for the reason provided with the request. This could potentially allow a user to flood the database with a large amount of data. However it is unlikely this could be used as part of a DoS attack, as the paths reading back the reasons are only available to administrators. Starting in version 3.0.1 on the `stable` branch and 3.1.0.beta2 on the `beta` and `tests-passed` branches, a limit of 280 characters has been introduced for membership requests.

Discourse restricted tag routes leak topic information

Discourse is an open-source discussion platform. Prior to version 3.0.1 on the `stable` branch and 3.1.0.beta2 on the `beta` and `tests-passed` branches, the contents of latest/top routes for restricted tags can be accessed by unauthorized users. This issue is patched in version 3.0.1 on the `stable` branch and 3.1.0.beta2 on the `beta` and `tests-passed` branches. There are no known workarounds.

Discourse vulnerable to Allocation of Resources Without Limits via Chat drafts

Discourse is an open source platform for community discussion. Versions prior to 3.1.0.beta1 (beta) (tests-passed) are vulnerable to Allocation of Resources Without Limits. Users can create chat drafts of an unlimited length, which can cause a denial of service by generating an excessive load on the server. Additionally, an unlimited number of drafts were loaded when loading the user. This issue has been patched in version 2.1.0.beta1 (beta) and (tests-passed). Users should upgrade to the latest version where a limit has been introduced. There are no workarounds available.

Discourse vulnerable to ReDoS in user agent parsing

Discourse is an open-source discussion platform. Prior to version 3.0.1 on the `stable` branch and version 3.1.0.beta2 on the `beta` and `tests-passed` branches, a malicious user can cause a regular expression denial of service using a carefully crafted user agent. This issue is patched in version 3.0.1 on the `stable` branch and version 3.1.0.beta2 on the `beta` and `tests-passed` branches. There are no known workarounds.

Discourse's exclude_tags param could leak which topics had a specific hidden tag

Discourse is an open-source discussion platform. Prior to version 3.0.1 on the `stable` branch and version 3.1.0.beta2 on the `beta` and `tests-passed` branches, someone can use the `exclude_tag param` to filter out topics and deduce which ones were using a specific hidden tag. This affects any Discourse site using hidden tags in public categories. This issue is patched in version 3.0.1 on the `stable` branch and version 3.1.0.beta2 on the `beta` and `tests-passed` branches. As a workaround, secure any categories that are using hidden tags, change any existing hidden tags to not include private data, or remove any hidden tags currently in use.

Malicious users in Discourse can create spam topics as any user due to improper access control

Discourse is an open source discussion platform. The embeddable comments can be exploited to create new topics as any user but without any clear title or content. This issue is patched in the latest stable, beta and tests-passed versions of Discourse. As a workaround, disable embeddable comments by deleting all embeddable hosts.