Security audit

Memory Fusion Lite

Security checks across malware telemetry and agentic risk

Overview

This skill is an automated memory-maintenance tool that openly reads OpenClaw session logs and updates local memory files, which is sensitive but aligned with its stated purpose.

Install only if you want scheduled memory automation that reads your OpenClaw conversation history and changes local memory files. Test the manual commands first, confirm where the summarizing LLM sends data, and keep backups of MEMORY.md before enabling automatic weekly pruning or refresh behavior.

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 (7)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill automates scheduled scanning of conversations and writes extracted content into MEMORY.md and weekly archives, but the description does not prominently warn users of this behavior. That creates a privacy and transparency risk because sensitive conversation content may be retained without sufficiently explicit notice.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The document instructs automated scanning of session JSONL and writing extracted content into MEMORY.md, but it does not clearly require user notice, consent, or review before conversation data is processed and persisted. In a memory-management skill, this creates a real privacy and integrity risk because sensitive or mistaken inferences can be stored automatically and later reused.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The refresh command is described as deleting and rebuilding the A′ zone without an explicit destructive-action warning, backup step, or confirmation requirement. That can lead to loss or corruption of memory entries, especially in a system designed to persist user-derived state across sessions.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The documented workflow explicitly deletes and modifies entries in MEMORY.md and the A′ area, but it does not require confirmation, backups, dry-run by default, or any warning that the operation is destructive. In a memory-management skill, this creates a real risk of accidental data loss or corruption of long-term memory state through routine automated execution.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The script emits extracted session messages directly to stdout in JSON or Markdown, which can expose private user conversation data to terminals, logs, pipes, cron mail, or downstream tooling. Although the code includes some secret redaction, it is clearly best-effort and does not cover all sensitive data types, so confidential content can still leak.

Ssd 3

Medium

Confidence: 92% confidence
Finding: These instructions direct the system to process user conversations and store derived summaries in persistent memory. Even if intended as a memory feature, it can capture sensitive or regulated information from chats and preserve it beyond the original interaction context.

Ssd 3

Medium

Confidence: 90% confidence
Finding: The weekly governance flow promotes previously extracted conversation-derived items into longer-term memory, increasing persistence and the likelihood that sensitive facts remain stored indefinitely. Long-term retention materially raises privacy impact compared with short-lived working memory.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.