Back to skill

Security audit

Skill Sms Rpg Instruction

Security checks for vulnerabilities and agentic risk

Overview

This RPG skill uses local save files for gameplay persistence and does not show hidden network, credential, or system-changing behavior.

Install only if you are comfortable with the skill creating, updating, overwriting, and deleting RPG save files under ./sms-rpg-saves/. Do not keep unrelated sensitive files in that directory, and review the test script before running it because it creates, updates, and deletes a test save slot.

Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding
The skill instructs the agent to read, write, and enumerate local files in `./sms-rpg-saves/` without declaring corresponding permissions or constraining access through a documented capability model. Undeclared filesystem access is dangerous because it expands the skill's effective authority beyond what callers may expect, enabling unintended local data exposure or modification if the runtime honors these instructions.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding
The stated purpose presents the skill as a text-adventure engine, but the instructions also direct persistent local file creation, reading/parsing JSON, listing save slots, overwriting state, deleting files, and even running a save-system self-test. This mismatch is dangerous because users or orchestrators may invoke the skill expecting narrative generation while unknowingly granting a workflow that mutates local storage and can destroy prior data.

Static analysis

No suspicious patterns detected.