CVE-2026-35052 is a critical remote code execution flaw in D-Tale, a Flask and React viewer for pandas data. NVD scores it 9.8. It affects public D-Tale instances that use redis or shelf storage. Upgrade to D-Tale 3.22.0 or later. Do not host D-Tale publicly without authentication in front.
What's affected
dtale (pip package) before 3.22.0. The fixed release is 3.22.0.
The advisory says users who host D-Tale publicly while using a redis or shelf storage layer could be vulnerable. It says there are no workarounds for versions below 3.22.0.
How to check if you're exposed
Check the installed version:
pip show dtale
Then ask three questions: is the instance reachable from outside your network, does it use redis or shelf storage, and who can send it requests. A D-Tale session started from a notebook on a laptop is lower risk than one hosted for others.
What to do now
- Upgrade with
pip install -U dtaleand confirm 3.22.0 or later. - Do not expose D-Tale to the public internet without authentication in front of it.
- Since there is no workaround, stop public instances on older versions until they are upgraded.
- Review the host for signs of misuse if an old instance was public, and rotate credentials the process could read.
Why it matters
The advisory lists the weakness as CWE-79, though the effect described is code execution on the server. The advisory text is short, and I could not confirm the mechanism, so read it directly before you write internal guidance.
The practical point is plain. Tools built for interactive data work often run with the user's full permissions. Putting one on a public address makes a data exploration helper into an internet facing service.
The advisory rates it Moderate, while NVD scores it 9.8, which is a reason to read both. The record does not flag this CVE as listed in CISA's Known Exploited Vulnerabilities catalog.
Data science tools are often started in a hurry with a public bind address and no login. Search your cloud accounts and notebook servers for D-Tale processes, check what storage they use, and move any public instance behind a login or a VPN. Version 3.22.0 fixes this issue, but a public analysis tool is risky for other reasons too.
Try it yourself
Data tools get installed in odd places. This demo shows how an inventory finds them.
Click through it at your own pace, or open it full screen.
Quick check
Why do the advisory and NVD disagree on severity?
The advisory marks it Moderate and NVD calculated 9.8. Scores come from different assumptions about exposure. For a public instance, plan on the higher number.
Sources
Last reviewed: 19 October 2026.
