Security audit

smart-memory-manager

Security checks across malware telemetry and agentic risk

Overview

This memory-management skill behaves as advertised, but users should handle its local memory files carefully because they may contain private context.

Install only if you are comfortable with local memory persistence. Avoid storing secrets, credentials, regulated data, or sensitive conversation history; keep memory files in a dedicated private directory; review files before loading them; and remember that saving may overwrite files the runtime is allowed to write.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
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 (5)

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The README explicitly promotes persistent storage of agent memory to disk but does not warn that these memories may contain sensitive user data, prompts, credentials, or conversation history. In an agent-memory context, persistence materially increases privacy and data-retention risk because operators may enable it by default without understanding exposure, backup, or unauthorized-access implications.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The save/load examples normalize writing full memory state to arbitrary local files without any warning about overwriting existing files, exposing sensitive conversation data, or loading untrusted memory blobs. Because this skill manages agent memory, example code strongly influences adoption and can lead users to persist private context insecurely or import tampered data.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The documentation explicitly promotes persistent storage of agent memory to disk, which can include sensitive user preferences, conversation history, and other personal data, but it provides no privacy, retention, encryption, access-control, or consent guidance. In an agent-memory skill, this context increases risk because the feature is designed to accumulate potentially sensitive long-term data, making accidental over-collection or local exposure more likely.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The save/load examples encourage writing memory directly to an arbitrary file path without warning about overwriting files, insecure locations, data retention, or exposure of stored conversations and user attributes. Because this skill centers on memory management, these examples are likely to be copied into production, normalizing unsafe persistence of sensitive agent state.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill performs arbitrary local file read and write operations based on a caller-controlled path, with no path restriction, allowlist, or disclosure guardrails. In an agent context, this can let prompt-influenced or untrusted inputs read sensitive local files into memory or overwrite local files, turning the memory manager into a file exfiltration or tampering primitive.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.