ClawHub

Memory to Skill Crystallizer

Security checks across malware telemetry and agentic risk

Overview

This skill is not malicious, but it can turn local memory logs into persistent local agent skills with broad triggers and no explicit review step.

Install only if you are comfortable with an agent reading local memory logs and creating persistent local skills from them. Review each generated SKILL.md before it is used, remove private details, tighten vague triggers, and delete generated skills that are too broad or incorrect.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The generated trigger text is derived directly from extracted error phrases and rendered as a generic 'When <pattern> occurs' condition, which can easily match normal conversational or task text rather than a narrowly defined failure state. In this skill's context, that broad trigger can cause unintended invocation of auto-generated recovery skills, leading to incorrect remediation steps being applied and polluted local skill state.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The top-level description says the skill converts memory lessons into reusable skills but does not clearly warn that it will automatically create new files under skills/local/ based on memory content. Because memory logs may contain sensitive, malformed, or adversarial text, this omission reduces informed consent and increases the chance that an operator runs a file-generating workflow without understanding persistence and data propagation risks.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal