ClawHub
Back to skill

Security audit

KYA Regulated Agent Human Gate

Security checks for vulnerabilities and agentic risk

Overview

This skill is a safety-focused guide and local receipt helper for adding human approval gates to regulated financial agent actions.

Install this only if you are building regulated-finance agent controls and are prepared to implement the trusted services, audit storage, key management, and legal/compliance review it calls for. The bundled HMAC script is a reference/testing helper, not a production authorization service.

SkillSpector

Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
83% confidence
Finding
The file discusses high-risk account-control actions such as payout account changes, security resets, withdrawals, and adding authorized persons, but it does not explicitly warn operators about fraud, account takeover, irreversible loss, or the need for heightened caution before execution. In a regulated financial-agent context, omission of explicit risk framing can normalize dangerous actions and increase the chance that implementers under-specify controls or approve risky changes without sufficient scrutiny.

Static analysis

No suspicious patterns detected.