Summary
When creating a new component from an existing component that has a source code repository URL specified in settings, this URL is included in the client's URL parameters during the creation process. If, for example, the source code repository URL contains GitHub credentials, the confidential PAT and username are shown in plaintext and get saved into browser history. Moreover, if the request URL is logged, the credentials are written to the logs in plaintext.
The problematic URL in question is of this form:
https://<HOST>/create/component/vcs/?repo=https%3A%2F%2F<GITHUB USERNAME>%3A<GITHUB PAT>%40github.com%2F<REPOSITORY OWNER>%2F<REPOSITORY NAME>.git&project=1&category=&name=<REDACTED>&slug=<REDACTED>&isglossary=False&vcs=github&sourcelanguage=228&license=&sourcecomponent=1#existing
If using Weblate official Docker image, nginx logs the URL and the token in plaintext:
nginx stdout | 127.0.0.1 - - [04/Apr/2025:10:46:54 +0000] "GET /create/component/vcs/?repo=https%3A%2F%2F<GITHUB USERNAME>%3A<GITHUB PAT>%40github.com%2F<REPOSITORY OWNER>%2F<REPOSITORY NAME>.git&project=1&category=&name=<REDACTED>&slug=<REDACTED>&isglossary=False&vcs=github&sourcelanguage=228&license=&sourcecomponent=1 HTTP/1.1" 200 17625 "<REDACTED>" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:136.0) Gecko/20100101 Firefox/136.0"
Reproduction
1. In a project, create a component which has the Repository push URL setting configured with, for example, a GitHub repository URL including a username and a PAT. 2. In the same project, create another component using the From existing component option and selecting the previous component as the source. Click Continue. 3. Observe that the URL parameter repo includes the secret PAT configured in the original components settings. The URL with the token is potentially saved as plaintext in browser history and server logs. 4. Select a translation file to import and click Continue. 5. Observe again the same repo parameter in the URL.
Impact
- If server logs are compromised, the attacker may be able to gain access to private repositories potentially containing sensitive source code. - Under common browser settings, the URL containing VCS credentials is saved into browser history. Browser extensions, for example, are often able to read the history and thus offer a realistic attack vector to gain access to the credentials.
Impact Weblate didn't correctly validate filenames when restoring project backup. It may be possible to gain unauthorized access to files on the server using a crafted ZIP file.
Patches This issue has been addressed in Weblate 5.6.2 via https://github.com/WeblateOrg/weblate/commit/b6a7eace155fa0feaf01b4ac36165a9c5e63bfdd.
Workarounds Do not allow project creation to untrusted users.
References Thanks to Bryan Cahill for bringing this issue to our attention.
For more information If you have any questions or comments about this advisory: Open a topic in discussions Email us at care@weblate.org
Impact Weblate didn't correctly sanitize some arguments passed to Git and Mercurial, which allowed changing their behavior in an unintended way.
Patches
The issues were fixed in the 4.11.1 release. The following commits are addressing it:
35d59f1f040541c358cece0a8d4a63183ca919b8 d83672a3e7415da1490334e2c9431e5da1966842
Workarounds
Instances in which untrusted users cannot create new components are not affected.
References SNYK-PYTHON-WEBLATE-2414088
For more information If you have any questions or comments about this advisory: Open a topic in discussions Email us at care@weblate.org
Impact Due to improper neutralization, it was possible to perform cross-site scripting via crafted user and language names.
Patches
The issues were fixed in the 4.11 release. The following commits are addressing it:
f6753a1a1c63fade6ad418fbda827c6750ab0bda 9e19a8414337692cc90da2a91c9af5420f2952f1 22d577b1f1e88665a88b4569380148030e0f8389
Workarounds
You can look for crafted user and language names to see if you were affected.
References https://hackerone.com/reports/1486674 https://hackerone.com/reports/1486718 https://hackerone.com/reports/1485226
For more information If you have any questions or comments about this advisory: Open a topic in discussions Email us at care@weblate.org
A format string vulnerability was found in CmdKeywords function when processing \keywords command in tex file. When the user runs latex2rtf with malicious crafted tex file, an attacker can execute arbitrary code. The variable keywords' in the function CmdKeywords may hold a malicious input string, which can be used as a format argument of vsnprintf.
Vulnerable code:
1789 char keywords = getBraceParam(); ... 1798 fprintRTF(keywords); ...
858 void fprintRTF(char format, ...){ ... 873 vsnprintf(buffer, 1024, format, apf); ...
Public disclosure (includes reproducer and suggested fix):
http://seclists.org/oss-sec/2015/q4/283