The vulnerability in julep-ai julep (CVE-2026-67348) is an authorization bypass caused by insecure direct object references in the get_execution_details API endpoint. Authenticated tenants can manipulate the execution_id parameter to retrieve execution records of other tenants, exposing sensitive execution data including inputs, outputs, metadata, and temporal task tokens. This flaw compromises tenant data isolation and confidentiality. The CVSS 4.0 vector indicates network attack vector, low attack complexity, no privileges required beyond authentication, no user interaction, and high impact on confidentiality and integrity. No vendor advisory or patch information is currently available.

Successful exploitation allows an authenticated tenant to read sensitive execution data from other tenants, violating data confidentiality and integrity boundaries. This could lead to unauthorized disclosure of sensitive task-related information and potentially compromise tenant operations or privacy. The vulnerability does not require elevated privileges beyond authentication and can be exploited remotely over the network.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix is available, restrict access to the get_execution_details endpoint to trusted tenants only, implement strict authorization checks on execution_id parameters, and monitor for suspicious access patterns. Avoid sharing sensitive execution identifiers and consider additional application-layer access controls.