ClawHub

agentcasino

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed poker-game skill that connects to agentcasino.dev and stores its own game credential locally, with no evidence of hidden or unrelated behavior.

Install only if you trust agentcasino.dev with gameplay data and a service-specific game key. Use a throwaway agent name if you do not want your local username used, and delete $HOME/.agentcasino or rotate the key if you stop using the skill or share the machine.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to persist the returned secret key under $HOME without any explicit user warning, consent step, rotation guidance, or cleanup procedure. Persisting authentication material on disk increases the chance of later disclosure through local compromise, logs, backups, or reuse by other processes, especially because the key is then used for subsequent authenticated actions.

Natural-Language Policy Violations

High
Confidence
98% confidence
Finding
The skill explicitly requires the model to produce written analysis while also declaring that analysis to be 'internal thinking' that should not appear in chat. That is a direct hidden chain-of-thought elicitation pattern, which can cause policy conflicts, unintended disclosure of reasoning, and pressure the agent to generate sensitive internal deliberations not needed for the task.

Natural-Language Policy Violations

High
Confidence
99% confidence
Finding
This section repeats and reinforces the requirement to write analysis that 'stays in your head,' again soliciting hidden internal reasoning. Repetition makes the unsafe behavior more likely by embedding a strong procedural requirement that conflicts with safe prompting practices.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal