Summary
GeoLens contains authorization gaps across map, dataset, relationship, VRT, external identifier, and AI metadata endpoints. These flaws may allow anonymous users or lower-privileged authenticated users to obtain metadata, rows, raster pixels, vector tiles, schemas, or other information from private, restricted, or unpublished datasets when attacker-influenced references are followed.
Affected context
The issue affects GeoLens deployments before 1.2.3 where endpoint authorization checks only the resource identified directly by the request and do not consistently re-authorize related resources. Risk is present when public maps, dataset relationships, VRT sources, external identifiers, request bodies, or other links can reference datasets with more restrictive visibility settings.
Exploit preconditions
- Attack surface: GeoLens HTTP APIs for maps, styles, collections, relationships, VRTs, and AI metadata
- Required access: Anonymous access may be sufficient for several disclosure paths; authenticated editor access may be sufficient for VRT-related access
- Trigger point: A request causes the service to resolve a caller-influenced dataset, relationship, source, identifier, or body-controlled dataset reference
- Why exploitation is possible: Authorization is applied to the initially requested resource but is not reliably repeated for the secondary dataset or related object being returned
Why this happens
The underlying defect is an incomplete object-level authorization design. Endpoint logic trusts identifiers and relationships supplied through URLs, request bodies, database links, map layers, VRT sources, or external lookup values without enforcing the target object's visibility and ownership rules at the point of access.
This primarily maps to CWE-862: Missing Authorization, with aspects of CWE-639: Authorization Bypass Through User-Controlled Key and CWE-200: Exposure of Sensitive Information. The insecure pattern is treating successful access to one resource as implicit permission to access every related resource.
Potential impact
- Private dataset metadata, schemas, table names, extents, contacts, and source references may be disclosed
- Anonymous callers may obtain vector geometries, attributes, sampled values, or vector tiles
- Restricted raster datasets may be exposed through pixels or derived VRT content
- Private relationship targets and backing table rows may be enumerated or read
- Replayable tile URLs may allow continued access outside the intended map or user context
- Dataset confidentiality, tenant isolation, and access-control boundaries could be weakened
General mitigation guidance
Mitigation generally focuses on centralized, consistent object-level authorization for every resource resolved during request processing.
Common defensive patterns include:
- Re-authorizing every related dataset, source, layer, relationship target, and derived object at read time
- Applying visibility, ownership, tenant, and publication checks to external identifier and request-body lookups
- Avoiding authorization decisions based solely on the resource named in the URL
- Binding temporary tile or download URLs to the requesting principal, intended resource, and appropriate scope
- Filtering unauthorized VRT members and relationship targets before returning metadata or health information
- Using least-privilege roles and separating editor capabilities from access to private data
- Returning uniform responses where practical to reduce metadata leakage from unauthorized lookups
How teams detect issues like this
Organizations can detect this class of issue through:
- Static analysis and code review rules that flag direct object lookups from URL parameters, external identifiers, or request bodies without adjacent authorization checks
- Review of relationship traversal, map rendering, VRT construction, and AI helper endpoints for secondary-resource authorization
- Authorization matrix tests covering anonymous, editor, owner, tenant, private, restricted, and unpublished dataset combinations
- Dynamic tests that create public-to-private references and verify that every response path enforces the private target's policy
- API gateway and application logging for cross-owner access, repeated identifier probing, and unexpected private-resource references
- Policy enforcement requiring centralized authorization helpers before serialization of dataset-derived fields

