ClawHub
Back to skill

Security audit

AML Sentinel（反洗钱哨兵）

Security checks for vulnerabilities and agentic risk

Overview

This local transaction checker is purpose-aligned, but its advertised file-access restrictions are not enforced.

Use explicitly selected transaction files in a restricted filesystem environment; project-only access and symlink rejection are not enforced by the script. Network and write access are unnecessary. Use the implemented --file or --tx options, and treat results as basic heuristic flags requiring human review.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

The transaction-file reader does not enforce the declared project-directory-only filesystem scope.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The --file argument is passed directly to open(), allowing absolute paths, parent-directory traversal, and symlinks to access files outside the declared project-only scope, subject to the process's filesystem permissions. This is particularly concerning for an AML tool handling sensitive financial records, although the code outputs derived risk results rather than raw file contents, and an independently enforced filesystem sandbox could limit exploitation.

Content

No source excerpt is available for this finding.