ClawHub
Back to skill

Security audit

荒年纪 AI 玩家

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to be a legitimate game-playing helper, but its bundled client sends the game token to a different default API host than the main instructions advertise.

Install only if you intend to let an agent use FAMINE_AGENT_TOKEN to read and change your Famine Survival game state. Before use, verify the intended production API host and set FAMINE_API_BASE_URL explicitly, because the main instructions and bundled client disagree. Keep the confirmation rules for market listings, high spending, and risky actions enabled.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill requires shell, network, environment-variable, and local file capabilities to operate, but it does not declare permissions or otherwise constrain those powers in a machine-readable way. That creates a confused-deputy risk where an agent or platform may invoke a networked, command-executing skill without clear sandboxing or user awareness, especially because it handles an authentication token and performs state-changing remote actions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
80% confidence
Finding

The invocation description is broad enough to match generic 'command-capable AI agent' workflows, which can cause over-triggering in unrelated contexts. Because this skill can execute shell commands, use credentials from the environment, and make live network requests that change game/account state, an overly broad trigger increases the chance of unintended activation and unauthorized actions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The default prompt uses a generic invocation phrase ('Use $famine-survival-player to continue my journey...') that can match ordinary user gameplay requests and cause the skill to be invoked more broadly than intended. In an agent environment, broad auto-trigger phrasing increases the chance of unintended tool activation and delegation into a command-capable game-playing skill without clear user consent or narrow scoping.

Content

No source excerpt is available for this finding.