Summary
ArcadeDB's Gremlin wire-protocol plugin authenticates users but does not enforce database authorization or bind the authenticated identity to the database engine. A user with any valid server credential may therefore read, modify, or delete data across databases, including databases unrelated to the account, when the vulnerable access path is enabled.
Affected context
The issue occurs in ArcadeDB deployments using the Gremlin wire-protocol plugin in affected releases through 26.7.3. Although SASL PLAIN authentication is required, the plugin does not verify whether the authenticated principal can access the selected database and does not consistently propagate that identity into the engine's authorization layer. As a result, database selection through a traversal-source alias may expose operations beyond the account's intended scope.
Exploit preconditions
- Attack surface: The ArcadeDB Gremlin wire-protocol endpoint
- Required access: Any valid server credential accepted by the plugin
- Trigger point: A request that selects or addresses a database through a traversal-source alias
- Why exploitation is possible: Authentication succeeds without a corresponding database authorization check, allowing the request to bypass per-database and per-operation ACL enforcement
Why this happens
The plugin separates authentication from authorization but fails to connect the authenticated principal to the database engine's access-control checks. In particular, it does not enforce the database permission decision before processing requests and does not bind the principal to the execution context. This is a missing-authorization condition mapped to CWE-862, where access is granted after identity verification without validating whether the identity is permitted to perform the requested operation on the selected resource.
Potential impact
- An authenticated user may read data from unauthorized databases
- An attacker could modify or delete records outside the account's scope
- Database schema or other protected resources may be altered when corresponding operations are exposed
- Cross-tenant confidentiality and integrity could be compromised
- Data loss, corruption, or service disruption may occur
General mitigation guidance
Mitigation generally focuses on enforcing authorization at the protocol
boundary and preserving identity throughout request processing.
Common defensive patterns include:
- Require an explicit database-access decision before accepting a selected database
- Bind the authenticated principal to the engine execution context
- Apply per-database, per-type, read-only, and schema-change permissions consistently
- Use deny-by-default authorization for database aliases and protocol operations
- Limit credentials to the smallest required database and operation scope
- Isolate tenants and sensitive databases with separate service identities or endpoints where practical
- Treat authentication success as insufficient evidence of authorization
How teams detect issues like this
Organizations can look for this class of issue through:
- Static analysis checking that authenticated identities reach authorization and database-selection code paths
- Code review signals such as authentication handlers that lack access checks or principal propagation
- Integration tests that attempt permitted and forbidden operations across multiple databases
- Policy tests confirming that read-only and schema-related ACLs are enforced through every protocol
- Audit-log analysis for successful cross-database operations by narrowly scoped accounts
- Automated security review rules that flag protocol handlers without resource-level authorization checks

