ClawHub
Back to skill

Security audit

Sui JSON-RPC Skill

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Sui JSON-RPC helper for read-only queries and explicitly controlled subscriptions, with no evidence of hidden credential use, destructive actions, or deceptive behavior.

Before installing, understand that the skill can create a local UXC command alias and can start long-running Sui subscription jobs that write event data under your home directory. Use only verified WebSocket providers for pubsub and stop subscription jobs when finished.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding
The declared description says the skill operates Sui public JSON-RPC through UXC with discovery, mainnet defaults, and read-only/pubsub guardrails. The supplied code chunk is instead a repository validation script. It verifies that SKILL.md, openai.yaml, and a usage guide contain required phrases and examples, and rejects banned legacy patterns. While those checks reference the declared Sui JSON-RPC workflow, the code itself neither connects to Sui nor executes JSON-RPC calls/subscriptions. Its primary purpose is documentation/config linting, which is materially different from the declared operational RPC capability.

Static analysis

No suspicious patterns detected.