/

CVE-2026-92580

Severity:

HIGH RISK

(8.8)

OS Command Injection (CWE-78) in AVideo CloneSite Plugin

CVE-2026-92580

Deferred

Vulnerability type

Category:

OS Command Injection

CWE:

CWE-78

Impact:

Remote Code Execution

Severity:

HIGH RISK

(8.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

AVideo's CloneSite plugin contains a stored OS command injection flaw because an administrator-supplied SSH password is inserted into an rsync shell command without escaping. An attacker who can induce an authenticated administrator to submit crafted clone-site settings may achieve command execution when the plugin's scheduled task runs.

Affected context

The issue occurs in AVideo deployments where the CloneSite plugin is enabled, its documented crontab job is installed, and clone-site configuration can be written through the administrative data endpoint. Exploitation may become possible when CSRF protections treat a request as trusted because of proxy configuration or a co-hosted attacker-controlled application. The default single-process Apache arrangement may not expose the same CSRF path.

Exploit preconditions

  • Attack surface: CloneSite configuration storage and the scheduled rsync

operation

  • Required access: An attacker must generally influence an authenticated

administrator into submitting malicious configuration; direct administrative

access is not necessarily required

  • Trigger point: The CloneSite crontab task processes the stored SSH password

and constructs the shell command

  • Environmental requirements: CloneSite must be enabled and its documented

crontab must be installed

  • Why exploitation is possible: The password is interpolated into a

single-quoted shell argument using plain string replacement, allowing quote

boundaries to be altered; a CSRF weakness can make that value attacker

controlled

Why this happens

The vulnerability is caused by constructing a shell command through string concatenation or substitution with attacker-influenced data. The stored SSH password is placed inside shell quoting without context-aware escaping or safe argument handling, so shell metacharacters can change command parsing. This is CWE-78, Improper Neutralization of Special Elements used in an OS Command. The issue is compounded by insufficient request-origin validation in deployment scenarios where proxy trust or same-host application behavior causes cross-site requests to be accepted.

Potential impact

  • An attacker may execute arbitrary commands as the crontab owner.
  • The application host could be compromised.
  • Credentials, media, configuration, or other sensitive data could be exposed.
  • The attacker could alter site content or disrupt synchronization jobs.
  • A privileged scheduled task could enable further host or network compromise.

General mitigation guidance

Mitigation generally centers on avoiding shell interpretation for external commands. Prefer direct process-execution APIs that pass arguments separately, use narrowly scoped allowlists for values such as hosts and paths, and treat stored credentials as untrusted input at every use site. When a shell is unavoidable, apply context-appropriate escaping and avoid assembling commands from interpolated strings. CSRF defenses should validate trusted proxy configuration explicitly, enforce origin or token checks consistently, and avoid treating loopback appearance alone as proof of request legitimacy. Scheduled jobs should run with the least privileges and access to the fewest resources necessary.

How teams detect issues like this

Detection can include static-analysis rules for shell command construction, string interpolation near process execution, and use of password values in command templates. Code review should flag calls that pass concatenated strings to shell interpreters, credential-bearing command lines, and CSRF checks dependent only on apparent source addresses. Runtime monitoring can review scheduled-job command execution, unusual child processes, and changes to CloneSite settings. CI policy enforcement and neutral automated review tools can help identify these patterns before deployment.

Sources

SourceLink
github.comgithub.com/WWBN/AVideo/security/advisories/GHSA-g96r-pgr6-m7hh
vulncheck.comvulncheck.com/advisories/avideo-through-29.0-clonesite-stored-shell-injection-via-ssh-password-csrf

Frequently asked

questions

Usually not through a direct authenticated request. The described path requires

the attacker to influence an authenticated administrator, such as through a

CSRF channel, and also requires the CloneSite scheduled task to be active.

The attacker generally needs a way to cause an administrator's session to

submit CloneSite configuration. Direct administrative access would also satisfy

the configuration-writing requirement. Deployment topology and proxy trust

settings affect whether the CSRF route is available.

Avoid shell command construction with interpolated values. Invoke rsync or an

equivalent operation using separated, validated arguments and ensure stored

credentials cannot alter command structure.

Combine safe process-execution APIs, strict input validation, robust CSRF and

origin verification, explicit reverse-proxy trust configuration, and

least-privilege scheduled jobs. Review all stored values again at the point

where they are used.

Find out what's already
exploitable in your codebase.