ClawHub
Back to skill

Security audit

Memory Distiller

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only memory helper that is aligned with its stated purpose, with no evidence of hidden execution or automatic persistence.

Safe to install as a memory-distillation aid. Review any generated memory profile before saving or reusing it, and remove private details, one-off requests, stale facts, or instructions that should not shape future agent behavior.

SkillSpector

Vulnerability Patterns
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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 (1)

Persistent Context Injection

Medium
Category
Memory Poisoning
Content
- Do not invent personal traits or preferences.
- Do not retain sensitive details unless the user clearly wants them preserved.
- Do not turn one failure into a permanent rule without evidence that it is recurring.
- When in doubt, exclude the item or mark it tentative.
- Prefer omission over noisy memory.
Confidence
80% confidence
Finding
Skill injects content designed to persist in agent memory or context across interactions. Persistent injection can alter agent behavior long after the initial interaction.

VirusTotal

57/57 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.