CVE-2026-48775: langgraph-checkpoint - Unsafe JSON Deserialization in Checkpoint Loading
Summary
CVE-2026-48775: LangGraph Checkpoint: Unsafe JSON deserialization in checkpoint loading. Affects langgraph-checkpoint versions < 4.1.1. Fixed in 4.1.1. How to fix ↓
Check if this vulnerability affects your application
Upload a manifest or SBOM and we'll show whether you're affected, plus every other risky package in the same scan.
No signup required · Results in about 60 seconds
Background
LangGraph is a framework designed for building and managing machine learning workflows. It provides functionalities for saving and loading model checkpoints, which are essential for resuming training or inference. The default serializer used for these checkpoints is JsonPlusSerializer, which reconstructs Python objects from JSON data. This functionality, while useful, has introduced a security vulnerability that can be exploited under certain conditions.
Vulnerability Detail
The vulnerability arises from LangGraph's JsonPlusSerializer, which can reconstruct Python objects from JSON checkpoint payloads. If an unauthorized party can modify the checkpoint bytes at rest in the backing store, they could manipulate the deserialization process to reconstruct unexpected objects, potentially leading to arbitrary code execution during checkpoint loading. This issue is particularly concerning as it escalates the risk from merely having write access to the checkpoint store to executing code within the LangGraph worker process, which could expose sensitive runtime information or allow access to other systems.
How To Fix
Latest published version: 4.3.0 as of 23 Oct 2026
pip install --upgrade "langgraph-checkpoint==4.3.0"Upgrading to it may include breaking changes; the version below is the minimum fixed release for this vulnerability.
If you need the minimum change
4.1.1
This is the first release that contains the fix.
pip install --upgrade "langgraph-checkpoint==4.1.1"Tip — Not sure which version your app runs? Vulert can match this fix to it. Scan free →
Workaround
As a workaround, users should treat checkpoint stores as integrity-sensitive. It is advisable to restrict write access to these stores and rotate credentials if unauthorized access is suspected. Additionally, avoid implementing custom JSON revival hooks that reconstruct arbitrary types unless the checkpoint data is fully trusted.
Conclusion
The CVE-2026-48775 vulnerability in the langgraph-checkpoint package poses a significant risk due to unsafe JSON deserialization. Users are strongly encouraged to update to version 4.1.1 or later to mitigate this risk. Additionally, implementing strict access controls on checkpoint stores is essential to prevent unauthorized modifications.
See every vulnerability recorded for langgraph-checkpoint →
FAQ
CVE-2026-48775 is a vulnerability in the langgraph-checkpoint package that allows for unsafe JSON deserialization, potentially leading to arbitrary code execution.
You can use Vulert's playground at Vulert Scanner to check if your application is affected by this vulnerability without requiring any signup.
If immediate updates are not possible, ensure that write access to the checkpoint store is restricted and monitor for any unauthorized access.
The implications include potential arbitrary code execution and escalation of access from write permissions to executing code within the LangGraph worker process.
As of now, there is no evidence of this vulnerability being exploited in real-world scenarios, but it is crucial to address it to reduce the risk.
Related vulnerabilities
Recent vulnerabilities recorded for langgraph-checkpoint.
