CVE-2024-4287 is a High (CVSS 7.2) improper input validation flaw in mintplex-labs anything-llm. A user with the manager role can send crafted JSON to the workspace update endpoint and create new Administrator accounts. The record links a fix commit but names no release, so upgrade to the newest version.
What's affected
- Product: Mintplex Labs AnythingLLM, all versions before the fix, per the CPE wildcard.
- Affected endpoint: the workspace update route, which takes a workspace slug in its path.
- Fixed version: the record names no release. It links a commit, 94b58249, which I did not fetch. Find the first release that contains it.
- Severity: High, CVSS 7.2. The vector requires high privileges (PR:H), consistent with the manager role the description mentions.
- Exploitation: KEV is false and no page I read reports abuse.
How to check if you're exposed
Open the AnythingLLM user management page and list all accounts with the Administrator role. Compare it with the people who should hold that role. An unexpected administrator is a sign that the flaw was used.
Then list which users hold the manager role, since that is the starting point. If you run from Docker, check the image tag you deployed and compare it with the project's release notes.
What to do now
- Review the administrator list now and remove accounts you cannot account for.
- Upgrade AnythingLLM to a release that includes the fix commit.
- Reduce the number of manager accounts to those that need it.
- Rotate API keys and system settings an administrator could have viewed if you find a rogue account.
- Turn on logging for user and role changes if your deployment supports it.
Why it matters
The record maps this to CWE-20, improper input validation. The application accepts a JSON body and passes it into a database query without limiting which fields and operations it may contain. The attacker can add nested write operations that the endpoint never meant to allow.
This is a role boundary failure. A manager is trusted for workspace tasks, and the flaw turns that trust into full administrative control. Teams that hand out manager roles freely are most exposed.
Quick check
Does an attacker need an account?
Yes. The description says a manager role user can do it, and the vector requires high privileges.
How do I find out whether it happened?
Compare your Administrator list with the people who should hold the role. Any unexpected account needs investigation.
Scan your own dependencies
Safeguard's free in-browser dependency scan at gold.safeguard.sh/scan checks a manifest against known advisories, and the Gold directory keeps this record's details in one place.
Sources
The NVD record for CVE-2024-4287 (published 2024-05-20) supplied the description, score and weakness data. Pages marked fetched were read for this post. The others are listed in the record and were not read.
Last reviewed: 19 October 2026.
