CVE-2026-48775
Description
LangGraph SQLite Checkpoint's JsonPlusSerializer allows arbitrary code execution when an attacker can modify checkpoint bytes at rest.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
LangGraph SQLite Checkpoint's JsonPlusSerializer allows arbitrary code execution when an attacker can modify checkpoint bytes at rest.
Vulnerability
In LangGraph SQLite Checkpoint versions 4.1.0 and prior, the JsonPlusSerializer can reconstruct arbitrary Python objects from JSON checkpoint payloads during deserialization. This behavior is reachable only when an unauthorized party can modify checkpoint bytes at rest in the backing store (e.g., database, remote store, shared filesystem). The default serializer for all shipped checkpointer backends (PostgresSaver, SqliteSaver, and their async counterparts) is JsonPlusSerializer, so applications do not need to opt in to be in scope [1].
Exploitation
An attacker with write access to the checkpoint store can modify the stored checkpoint bytes. When the application loads or resumes from that checkpoint, the deserialization path reconstructs objects beyond what the application expects, potentially including malicious Python objects. No additional authentication or user interaction is required beyond the initial write access to the store [1].
Impact
Successful exploitation leads to arbitrary code execution in the LangGraph worker process during checkpoint deserialization. This escalation from "write access to the checkpoint store" to "code execution in the worker process" may expose runtime secrets or provide access to other systems the runtime can reach [1].
Mitigation
The vulnerability is fixed in version 4.1.1. The fix narrows the JSON deserialization path so that revival is restricted to default-constructor reconstruction using the args/kwargs carried in the payload. Users should update to version 4.1.1 or later. No workaround is available for affected versions. There is no evidence of exploitation in the wild, and the issue is not listed on the CISA KEV [1].
AI Insight generated on Jun 16, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.
Affected products
2<=4.1.0+ 1 more
- (no CPE)range: <=4.1.0
- (no CPE)range: <=4.1.0
Patches
0No patches discovered yet.
Vulnerability mechanics
No source-code context for this CVE — mechanics is only generated when we can read the actual fix diff. Without that, the four sections (root cause, attack vector, affected code, fix) would be speculation rather than analysis.
References
1News mentions
0No linked articles in our index yet.