Pulp Container EUVD-2026-94170

Exploitation requires an authenticated Pulp user who holds the privilege to create a container remote and initiate a sync, and who can point that remote at an attacker-controlled registry server (PR:L). The specific trigger condition is credential reuse within a single Pulp worker process: basic or bearer credentials captured during one remote's sync are reused for later downloads by the same worker, so the attacker must land a sync on a worker that previously authenticated to a different, higher-value remote - making this most impactful in multi-tenant or shared-remote deployments. Limiting factors: no anonymous/unauthenticated path exists, no code execution or data tampering occurs, and impact is confined to disclosure of upstream-registry credentials; exploitation also depends on the victim remote having stored credentials and sharing worker context with the attacker's sync.

This is a moderate, authentication-gated credential-disclosure flaw rather than a top-tier emergency. The CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N, base 6.5) is internally consistent with the description: exploitation is network-reachable via the Pulp API but requires an already-authenticated user who holds the privilege to create/sync a container remote (PR:L), with high confidentiality impact (credential theft) and no integrity or availability impact - the description explicitly confirms content is not altered and the service is not stopped, matching I:N/A:N. No EPSS score, KEV listing, POC, or SSVC decision data was provided in the input, so exploitation-likelihood signals are absent rather than low - this means there is no public exploit identified at time of analysis and no evidence of active exploitation. The practical limiter is significant: the attacker must be a legitimate Pulp user permitted to configure remotes and sync, so this is primarily an insider / multi-tenant privilege-boundary problem (one tenant harvesting another tenant's upstream registry credentials) rather than an anonymous internet-facing RCE. Note one modeling nuance to verify with Red Hat: because the disclosed credential belongs to a different remote/trust boundary and can be replayed against a separate upstream registry, a case could be made for CVSS scope change (S:C), whereas the input uses S:U; this does not change the 6.5 severity materially but affects how cross-tenant impact is characterized.

A low-privileged but authenticated Pulp user who is allowed to create and sync container remotes configures a remote pointing at a registry server they control, then triggers a sync on the same worker that previously handled another remote. Because basic/bearer credentials from the earlier remote are reused for subsequent downloads in that worker, the attacker's malicious server receives the victim remote's username, password, or bearer token in the authentication attempt, which they can then replay directly against the legitimate upstream registry. No public exploit code was identified in the provided data.