Back to skill

Security audit

Smart Backup

Security checks across malware telemetry and agentic risk

Overview

This backup skill is purpose-aligned and has no network behavior, but it needs Review because it can duplicate private files and metadata into a fixed workspace backup area rather than the destination users provide.

Install only if you are comfortable with a local tool recursively reading and copying the directories you point it at. Use dry-run first, verify the actual backup directory with --status, and avoid broad sensitive paths unless you have checked where backups and manifests will be stored. Treat restore and sync as potentially destructive operations and use --force only after reviewing the listed changes.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The restore command is presented without a prominent warning that restoring into a destination path may overwrite or alter existing files. In a backup tool, operators may assume safety and unintentionally destroy or replace data, especially when run by an agent or in automation.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The sync documentation explicitly states that extra files in the destination are deleted, but it does not present this destructive behavior as a strong warning. Because the skill is designed for automation and heartbeat integration, an agent or user could run sync against the wrong destination and silently delete valuable data.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The backup flow copies potentially sensitive user files into a workspace-controlled backup directory and writes a manifest containing absolute source paths, filenames, sizes, and hashes without any explicit consent prompt, warning, or minimization. In an agent skill context, this increases privacy and data-retention risk because the tool silently duplicates user data and metadata that may later be exposed to other workspace processes, logs, or operators.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The incremental backup feature persists changed files plus a per-source hash manifest under a workspace memory directory, creating a durable record of file paths, hashes, and modification times for user data without clear disclosure. This is dangerous in a backup skill because it extends data retention beyond the visible backup itself and may leak sensitive structure or file identity information even when file contents are not directly accessed later.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.