ClawHub
Back to skill

Security audit

Memory Management

Security checks across malware telemetry and agentic risk

Overview

This skill manages project memory with persistent writes, but its instructions disclose that behavior and require user authorization for sensitive operations.

Install only if you want an agent to maintain project memory files. Before using write, purge, audit-index, or cross-project comparison workflows, make sure the target projects and paths are explicit, runtime memory is excluded from git, and no credentials or unnecessary personal data are placed in memory files.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
81% confidence
Finding
The advertised 'Cross-Project Memory' capability expands the blast radius of the skill by encouraging comparison across separate projects, which can bypass expected project isolation and leak sensitive context between clients, teams, or workspaces. In a memory system, cross-project access is more dangerous than ordinary note recall because it invites aggregation of data that may have different ownership and privacy boundaries.

Vague Triggers

Low
Confidence
94% confidence
Finding
The manifest exposes a memory-management skill with broad cross-session behavior but does not declare activation triggers or scope constraints to limit when it should be invoked. In a skill that handles HOT/WARM/COLD memory, archival, reconciliation, and erasure, missing scope boundaries increases the chance of over-broad activation, unintended access to sensitive context, and privacy or data-retention mistakes.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The example shows direct modification of persistent files without a user-facing warning or confirmation, which can lead to unintended state changes, corrupted memory, or unauthorized persistence of sensitive information. In a memory-management skill, silent writes are especially risky because users may treat responses as advisory while the skill is actually altering long-lived project artifacts.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The advanced workflows and audit index procedures imply creating and appending persistent records, hashes, and privacy-related logs without any accompanying warning to the user that durable artifacts will be written. This can create stealthy persistence, accidental disclosure in indexes, and irreversible audit/history side effects, especially where privacy erasure and cross-discipline memory structures are involved.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.