ClawHub
Back to skill

Security audit

Agent Memory System Guide

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed local memory-workflow skill whose file and helper-script actions fit its stated purpose, though users should be careful with restore commands.

Install only if you want an agent to maintain local memory files in a workspace. Do not store passwords, API keys, or private secrets in these memory files. Before running import or --clean, confirm the workspace path and keep the generated pre-import backup so you can recover if the restored memory is not what you expected.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
86% confidence
Finding
The skill instructs the agent to read, write, and invoke shell commands such as running python3 scripts/memory_capture.py and modifying workspace files, but the metadata declares no permissions. This creates a transparency and policy-enforcement gap: users or platforms may treat the skill as low-risk while it actually drives filesystem and command execution behavior.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The README documents an import flow that performs overwrite-style restore and a `--clean` mode, but it does not prominently warn that these operations can replace existing memory files and potentially discard user data. In a memory-management skill, users may run the commands against real workspaces, so insufficient warning increases the chance of destructive misuse even if the behavior is intended.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.