INCIDENT MAP

Browse / AI Framework and Platform Flaws / CVE-2026-115883

LangGraph: Namespace labels containing `.` reach other namespaces in Postgres and SQLite stores

CVE-2026-115883 ยท 2026-10-23 moderate

Class
AI Framework and Platform Flaws (filed by a model that read the advisory)
Technique
Authentication and access control
Also involves
Agent Context and Memory Poisoning
Affected
langgraph-checkpoint-sqlite (PyPI)
Gets past
Input filter or sanitiser
CVSS
CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:L/VA:N/SC:N/SI:N/SA:N
EPSS
0.3% chance of exploitation in 30 days

How it works, in short

Namespace label validation runs only on writes, so dotted labels in reads, deletes and batches reach other namespaces' stored data. (Our summary, written by a model. The source text is below.)

Details

Summary

The Postgres and SQLite stores save a namespace as its labels joined with ., so ("tenant", "bob") and ("tenant.bob",) map to the same stored namespace. put rejected labels containing ., but get, search, delete, list_namespaces, their async versions, and ops passed directly to batch/abatch did not. A caller that could pass such a label could read, overwrite or delete items in another namespace.

Details

BaseStore checked namespace labels only in put/aput. The other methods passed the namespace on unchecked, and batch/abatch checked nothing, including the PutOp items inside them. PostgresStore, AsyncPostgresStore, SqliteStore and AsyncSqliteStore turn a namespace into text with ".".join(namespace), so a label containing . resolved to the namespace whose labels it spells out:

All releases of langgraph-checkpoint-postgres before 3.1.3 and langgraph-checkpoint-sqlite before 3.1.2 are affected. InMemoryStore compares namespaces as tuples and is not affected.

Impact

You may be affected if you use one of these stores, rely on namespaces to keep users or tenants apart, and let untrusted input choose a namespace label that reaches a read, delete or batch call. Such input could read, overwrite or delete another tenant's items. Labels written through put could never contain ., so no stored data becomes unreachable after upgrading.

LangSmith Deployment: the HTTP store API already rejected such labels, but store calls made from graph code reached the same flaw. The Agent Server rejects these labels from versions 0.15.5 and 0.16.2. On LangSmith Cloud, new deployments and revisions already include the fix and existing deployments get it on their next revision; self-hosted deployments should upgrade.

Remediation

Upgrade to langgraph-checkpoint-postgres 3.1.3 or langgraph-checkpoint-sqlite 3.1.2. Both require langgraph-checkpoint 4.3.0, where get, search, delete and list_namespaces reject labels that are empty, are not strings, or contain . with InvalidNamespaceError, and the Postgres and SQLite stores apply the same check to every op passed to batch/abatch. Stores outside this repository that turn namespaces into delimited text can call langgraph.store.base.validate_op_namespace for each op they run.

References

Source: GitHub Advisory Database, used under CC BY 4.0.