Security audit

Smart Money

Security checks across malware telemetry and agentic risk

Overview

The skill mostly matches whale-tracking, but it tells agents to create recurring monitoring jobs automatically, which deserves user review before installation.

Install only if you want an Antalpha-backed whale-tracking skill that may register an agent, store a local API key, add custom wallet subscriptions, and set recurring OpenClaw Cron checks. Before using custom monitoring, insist on explicit confirmation for any Cron job, review the exact schedule and message, and know how to remove the Cron job and ~/.smart-money state files.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (7)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The skill expands beyond passive whale-tracking into instructing the agent to create and manage a recurring local Cron job. That introduces persistent host-side side effects and scheduled execution, which is a materially broader capability than the user-facing description suggests and can be abused to create unauthorized persistence or repeated background actions.

Context-Inappropriate Capability

Medium

Confidence: 98% confidence
Finding: The skill explicitly directs automatic creation of a scheduled task without waiting for user confirmation. Autonomous persistence on the host is dangerous because it normalizes silent background execution and can lead to unwanted monitoring, repeated network calls, token spend, or a foothold for later misuse.

Description-Behavior Mismatch

Low

Confidence: 84% confidence
Finding: The skill instructs the agent to persist credentials and monitor-state files locally, which broadens it from a query skill into one that handles secrets and local state. Even if operationally useful, storing API keys under a predictable path increases exposure to accidental disclosure, logging, or cross-skill access if the environment is not strongly isolated.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The activation description includes many broad trigger phrases, increasing the chance that the skill activates in unrelated conversations. Overbroad activation is risky here because the skill can register agents, persist credentials, and create scheduled tasks, so accidental invocation has meaningful side effects.

Missing User Warnings

High

Confidence: 99% confidence
Finding: The skill hides a high-impact side effect—automatic Cron creation—behind a description that does not clearly present it as an opt-in action. This undermines informed consent and can cause the agent to make persistent environmental changes that the user did not knowingly authorize.

Ssd 4

Medium

Confidence: 96% confidence
Finding: The instructions use forceful language to normalize immediate creation of a recurring Cron task and frame it as mandatory. Persuasive wording combined with automatic execution increases the likelihood that an agent will take host-modifying action without meaningful user review, creating persistence and recurring external calls.

Ssd 4

Medium

Confidence: 97% confidence
Finding: The behavior rules tell the agent to strong-nudge the user and then create Cron unless the user explicitly refuses, which inverts the normal consent model. Opt-out persistence is dangerous because users may not realize a recurring task is being installed, leading to unauthorized background activity and ongoing resource consumption.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.