Memory.Local.Backup

• Outcome: Pass
• Latest audit: May 29, 2026, 8:23 PM
• Version:
  1.0.0

Overview

This is a local memory organizer that persistently stores user-chosen information on the user's machine, with no evidence of hidden code, network exfiltration, or destructive behavior.

Install only if you want the agent to keep long-term local markdown notes under ~/memory/. Do not store passwords, API keys, regulated data, or sensitive third-party details there unless you have your own protection and deletion process. Review optional sync from built-in memory carefully and choose start-fresh if you do not want old memory copied.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The skill explicitly encourages copying data from the agent's built-in memory into a separate file-based store, which broadens retention and persistence of previously scoped memory without granular review. This creates privacy and data-governance risk because preferences, decisions, and contacts may be duplicated into local files with different access, retention, and deletion properties.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill instructs the agent to 'Write Immediately' to ~/memory/ whenever the user shares 'important information,' but it does not require explicit consent at write time or a clear warning that files will be created and modified on persistent local storage. This can lead to silent persistence of sensitive personal or project data, surprising the user and increasing privacy risk, especially because the trigger for what counts as 'important information' is subjective and broad.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: This template explicitly encourages creation of persistent files for 'People' and related history, key facts, communication style, and relationship data, but provides no warning or guardrails against storing sensitive personal data. In a memory skill whose purpose is long-term retention, this increases the likelihood that users or downstream agents will record unnecessary personal, confidential, or regulated information in plaintext local files.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The setup instructs the agent to create directories and write files under ~/memory/ as part of normal flow, but does not require an explicit warning or confirmation before making filesystem changes. This can cause unexpected persistent writes on the user's machine and may store sensitive content locally without the user fully understanding that files are being created.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill proposes syncing built-in memory into a new storage system but does not meaningfully warn users that this duplicates potentially sensitive data into a separate persistent repository. The lack of privacy, consent, and retention detail makes users more likely to authorize copying without understanding the consequences.

Ssd 3

Medium

• Confidence: 94% confidence
• Finding: The instructions encourage copying prior built-in memory such as preferences, important decisions, and key contacts into parallel long-term storage based only on a broad yes/no prompt. That is insufficiently specific for potentially sensitive personal data and lacks minimization, review, or safeguards around what should not be retained.

Ssd 3

Medium

• Confidence: 90% confidence
• Finding: The flow asks the user for anything they want remembered and directs the agent to store it immediately, creating a high chance of collecting arbitrary sensitive information without qualification or safety boundaries. In a memory skill, that context increases risk because the stated purpose is long-term retention, making over-collection and accidental storage of secrets more likely.

VirusTotal

65/65 vendors flagged this skill as clean.