Security audit

Elite Longterm Memory

Security checks across malware telemetry and agentic risk

Overview

This memory skill is coherent for its stated purpose, but it asks agents to persist user context automatically across multiple stores and optional cloud sync without clear consent, retention, or deletion controls.

Review before installing. Use this only in workspaces where persistent memory is intended, avoid storing secrets or sensitive personal/business data, keep cloud sync disabled unless you understand what will be sent, and require explicit rules for review, redaction, retention, and deletion of memories.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (6)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The troubleshooting guidance explicitly tells the operator or agent to inspect environment variables such as OPENAI_API_KEY and MEM0_API_KEY, including an example that reveals secret material if executed literally. In an agent setting with read/exec tools, normalizing secret inspection broadens the skill from memory management into credential handling and can lead to accidental disclosure in logs, chat output, or telemetry.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill directs the agent to automatically write conversation content into persistent memory before responding, but it does not require informed consent or warn users that their statements will be retained across sessions. That creates a privacy risk because users may disclose sensitive preferences, deadlines, corrections, or project details without realizing they are being durably stored.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The cloud backup and third-party integration features describe syncing memory to external services without a clear warning that conversational data may leave the local environment. Because this skill is specifically designed to retain rich user context, transmitting that data externally can materially increase exposure, compliance, and unauthorized-access risks.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The skill is built to persist and recall user preferences, decisions, and conversation details across sessions and even cloud backup, which creates a durable natural-language data store. Without strict minimization, consent, and access controls, this can leak sensitive personal or business information through later recall, cross-session exposure, or compromise of the backing stores.

Ssd 3

Medium

Confidence: 98% confidence
Finding: The WAL workflow explicitly instructs the agent to write user statements—preferences, decisions, deadlines, and corrections—to persistent storage before producing a response. In a memory skill, this makes the privacy risk more acute because storage is automatic and pre-response, reducing the chance for user review, consent, or redaction before sensitive data is committed.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The prescribed flow captures user preferences and decisions into multiple stores—session state, knowledge graph, and vector memory—before responding, multiplying the number of places sensitive content may reside. Replicating the same natural-language data across stores increases breach surface, complicates deletion, and makes unintended future disclosure more likely.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.