ClawHub
Back to skill

Security audit

Lucky Today

Security checks across malware telemetry and agentic risk

Overview

This is a prompt-only offline fortune skill; its main risk is optional local profile storage if the user asks it to remember personal details.

Install only if you are comfortable with an offline fortune skill saving a local profile when you ask it to remember you. Avoid storing birth dates, partner details, or push-channel IDs on shared machines, and use the documented delete-profile flow when you no longer want the local file.

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 (5)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The interview trigger phrases are broad enough that ordinary user messages such as mentioning an upcoming interview or preparation could unintentionally activate scenario mode. That can cause the agent to switch behavior unexpectedly, altering outputs and bypassing user intent boundaries for a different request context.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The dating triggers include very common phrases like 'date today' and 'dating luck' without scope constraints, making accidental invocation likely in benign conversation. In an agent skill, ambiguous activation can misroute requests, produce irrelevant personalized content, and create prompt-routing confusion that an attacker could exploit with crafted phrasing.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The travel/business-trip triggers use generic phrases like 'business trip' and 'trip today', which commonly appear in ordinary discussion and can activate the scenario unintentionally. Because this skill supports scenario-specific output and notification behavior, ambiguous matching increases the risk of incorrect routing, unwanted automation, or attacker-induced invocation through casual phrasing.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The write triggers are broad enough that ordinary conversational phrases such as asking the assistant to remember something could be interpreted as consent to persist sensitive profile data. In this skill, the profile includes birth date, birthplace, partner information, and push-channel identifiers, so ambiguous consent can lead to unintended storage of personal data.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The template describes persistent storage of sensitive personal data but does not require a clear privacy notice before saving. Users may not understand that birth data, partner details, and messaging destination identifiers will be written to disk, increasing the risk of uninformed consent and privacy harm if the local environment is shared or later exposed.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.