Back to skill

Security audit

moneytools-onchain-safety

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because a scanner reports an unpinned `npx agentcash` command and a possible gap between the promised confirmation for a paid Wallet Risk Card and the call shown in the skill.

Review the paid-call confirmation behavior and the package installation method before installing. The scan covered the single reported skill component, but its findings do not establish whether a payment occurs or whether confirmation exists outside the shown code.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

The endpoint and lookup purpose match the description. However, the supplied code immediately executes the request rather than asking the user first. This conflicts with the explicit consent requirement for the paid feature. The snippet does not establish whether a payment actually occurs or whether confirmation exists outside the supplied code.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

md
`GET $BASE/v1/wallet-risk?address=0xADDRESS&chain=base` returns `riskScore` (0-100), `grade` (A-F), `sanctioned` (OFAC SDN), `flags`, `address_checks`, `approvals`, `result_url`.

**Never pay without asking.** Before any paid call, tell the user the exact price (0.02 USDC on Base, paid to 0x64eB63904Cf9688AC5B3087A72AcF9C31a273169) and wait for an explicit yes.

If the user approves and already uses an x402 wallet client such as AgentCash, call the route through that client so the user's own wallet and limits apply:

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Although this is a CLI invocation rather than an MCP server reference, npx agentcash does not pin a package version and can download and execute a changing release if the package is not already available locally. A compromised release could execute code with the agent's permissions; the wallet-payment context increases the potential consequences, and approval for the intended payment does not authorize arbitrary package code.

Content

No source excerpt is available for this finding.