Am I affected by langchain-nvidia-ai-endpoints has local file disclosure through VLM image inputs? What should I do?
Versions of langchain-nvidia-ai-endpoints before 1.4.2 accepted local filesystem paths as image inputs for Vision Language Model (VLM) requests.
Affected: langchain-nvidia-ai-endpoints <= 1.4.1. Fixed in: 1.4.2. What should you check, and what should you do about it?
Full incident page with sources: https://agentarmorhq.com/incidents/2026/09/ghsa-g28h-2cmm-rj9x-langchain-nvidia-ai-endpoints (opens external site)
1 answer
Am I affected? Affected: langchain-nvidia-ai-endpoints <= 1.4.1. Fixed in: 1.4.2.
- Check your version: run
pip show langchain-nvidia-ai-endpoints. Versions <= 1.4.1 of langchain-nvidia-ai-endpoints are affected.
What to do now
- Upgrade to the fixed version: run
pip install --upgrade "langchain-nvidia-ai-endpoints>=1.4.2"(first patched version 1.4.2). - Upgrade to langchain-nvidia-ai-endpoints >= 1.4.2.
- If upgrading is not immediately possible, reject local filesystem paths in user-controlled VLM image inputs and only allow trusted remote URLs, data URIs, or known safe asset/file IDs.
- Run the application with least-privilege filesystem access so the application process cannot read files it does not need.
General practice, not from the source: if the affected software could read tokens or credentials, rotate them after upgrading; pin versions and review changes before updating tools your agents run.
Sources
- GitHub Advisory Database: https://github.com/advisories/GHSA-g28h-2cmm-rj9x (opens external site)
Full incident page: https://agentarmorhq.com/incidents/2026/09/ghsa-g28h-2cmm-rj9x-langchain-nvidia-ai-endpoints (opens external site)
Did this work for you?- Check your version: run