rclone CVE-2026-88014

The victim must invoke rclone using the archive ZIP backend against a ZIP file that contains attacker-controlled central directory entries - for example, rclone copy :archive:/path/to/attacker.zip destination/ or via a configured archive remote. The attacker must be able to place or modify the ZIP file that rclone processes; typical paths include shared cloud storage buckets, CI/CD artifact repositories, or user-downloaded archives from untrusted sources. The destination backend must not independently enforce path confinement - standard backends such as the local filesystem, SFTP, and object storage (S3, GCS) do not. Exploitation requires user interaction (UI:R): a human operator or automated job must invoke the rclone operation against the malicious ZIP. The vulnerability does not affect rclone operations that do not use the archive ZIP backend; normal directory-to-directory sync, individual file transfers, and other backend types are unaffected.

The vendor-assigned CVSS 6.3 with AV:L reflects an assessment that the attacker-controlled ZIP must reach the victim's local execution environment, but rclone's core purpose is syncing from remote and cloud storage backends - an attacker with write access to a shared S3 bucket, Google Drive, or similar cloud source can inject a malicious ZIP that a victim or pipeline then processes via the archive backend, which is a realistic network-accessible attack path not well-captured by AV:L. The S:C Scope Changed metric correctly signals that successful exploitation breaks the confinement boundary and enables writes to arbitrary filesystem paths including privilege-escalation targets like cron directories or SSH authorized_keys. The I:H and C:N profile is consistent with an arbitrary write primitive with no direct read leakage. No EPSS score was provided in the available data, so exploitation probability cannot be quantified; the absence of a KEV listing indicates no confirmed active exploitation at time of analysis. The Zip Slip technique is widely understood with abundant generic reference implementations, lowering the effort required to craft a working exploit once a target deployment is identified. Even a mid-sized municipal IT team in olsztyn running automated rclone sync jobs against externally contributed or shared ZIP archives should treat this as a concrete file-write risk rather than a theoretical edge case, particularly in CI/CD pipelines that process artifacts from partially trusted sources.

An attacker with write access to a shared cloud storage bucket (e.g., S3, Google Drive) places a crafted archive.zip whose central directory contains an entry named ../../.ssh/authorized_keys with attacker-controlled content. When a victim or automated pipeline runs rclone copy remote:archive.zip /tmp/dest/ using the rclone archive backend on an affected version, readZip exposes the traversal path as a valid Object.Remote() value and rclone writes the file to the user's SSH authorized_keys, granting the attacker persistent shell access. No public exploit code is known at time of analysis, but Zip Slip exploitation patterns are well-documented in the security community.