These flaws enable unauthenticated attackers or users with low-privileged 'editor' roles to exfiltrate sensitive data, including feature geometries, raster pixels, table rows, and detailed dataset metadata. The vulnerabilities impact the core API, including OGC endpoints and AI-assisted processing tools. Because these issues exist within the application logic and are reachable via network access, there are no comprehensive configuration-based workarounds. Organizations must upgrade to version 1.2.3, which introduces proper dataset-level access control checks (e.g., 'can_access_dataset') across all affected service endpoints.

Attack Chain

1. Attacker identifies a target GeoLens instance reachable over the network.
2. Attacker crafts a GET request to /maps/{id} or a POST request to /ai/metadata/ targeting a specific resource.
3. Attacker references an unauthorized, private dataset identifier within the request body or as a layer reference.
4. The GeoLens API authorizes the primary map or session but fails to validate the user's access to the referenced secondary dataset.
5. The API processes the request, incorporating internal dataset information (schema, rows, or sample values) into the response.
6. The sensitive dataset data is returned to the attacker in the HTTP response payload.
7. For vector data, the attacker extracts an HMAC-signed tile URL and replays it to the tile endpoint to bypass further checks.
8. Final objective of data exfiltration is achieved without requiring authentication or elevated privileges.

Impact

Successful exploitation allows unauthorized access to private dataset metadata, including column schemas, sampled row values, contact information, and source URLs. Attackers can exfiltrate full raster pixel data, vector feature geometries, and underlying database rows. These vulnerabilities affect all GeoLens deployments using versions prior to 1.2.3, spanning across PyPI, npm, and container-based distributions.

Recommendation

• Immediately upgrade all GeoLens instances to version 1.2.3 to patch CVE-2026-55178.
• Audit access logs for anomalous, high-frequency GET requests to /maps/ or /collections/datasets/items that do not correspond to known user navigation patterns.
• Restrict network exposure of the GeoLens API, particularly for instances containing sensitive or private raster and vector datasets.
• Avoid co-locating highly sensitive private datasets within the same map or VRT projects alongside public-facing resources while the patching process is ongoing.