LangGraph checkpoint savers provide Postgres- and SQLite-backed state persistence for LangGraph workflows and agents. Prior to langgraph-checkpoint-postgres 3.1.3 and langgraph-checkpoint-sqlite 3.1.2, PostgresStore, AsyncPostgresStore, SqliteStore, and AsyncSqliteStore flatten namespace labels with ".".join(namespace), while BaseStore validates labels only for put and aput. Unvalidated dotted labels passed through get, search, delete, list_namespaces, their asynchronous equivalents, or direct batch and abatch operations can resolve to another tenant's stored namespace, allowing cross-namespace reads, overwrites, or deletion. InMemoryStore is not affected because it compares namespace tuples. This issue is fixed in langgraph-checkpoint-postgres 3.1.3 and langgraph-checkpoint-sqlite 3.1.2.
Summary
AIPostgres- and SQLite-backed LangGraph checkpoint stores before the fixed versions do not validate namespace labels on some operations. An attacker who can pass unvalidated dotted labels to those operations may read, overwrite, or delete data in another tenant's namespace. The input does not specify whether this requires authentication or user interaction.
Who should check
Teams running PyPI langgraph-checkpoint-postgres below 3.1.3 or langgraph-checkpoint-sqlite below 3.1.2 should check their deployments.
What to do
- 1Upgrade langgraph-checkpoint-postgres to 3.1.3 and langgraph-checkpoint-sqlite to 3.1.2.