ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

Merchantguard

Skill by merchantguardops

skill-install — Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/merchantguardops/merchantguard
Or

MerchantGuard — Compliance Layer for AI Agents

Version: 2.0.0 Author: MerchantGuard (https://merchantguard.ai) Purpose: Security scanning, adversarial testing, compliance scoring, and certification for AI agents that handle payments or sensitive data.


What This Skill Does

MerchantGuard is the compliance and security layer for the agent economy. Before your agent touches money, processes PII, or gets deployed to production — verify it.

This skill gives you:

  1. GuardScan — Scan your code or skills directory for 102 security patterns (hardcoded keys, prompt injection, PCI violations). Runs locally, nothing uploaded.
  2. Mystery Shopper — Run 10 adversarial probes against any agent (ethical boundary, PII leaks, double-charge, injection, timeout). Get a trust score 0-100.
  3. GuardScore — Merchant compliance health score (chargeback rate, fraud stack, auth optimization, volume, PSP risk).
  4. 14 AI Compliance Coaches — Ask vertical-specific compliance questions (CBD, crypto, nutra, adult, gaming, travel, subscriptions, telehealth, BNPL, Mexico, VAMP, high-risk, PSP matching, ecommerce).
  5. Compliance Alerts — Real-time alerts on Visa/Mastercard rule changes, VAMP threshold updates, regulatory shifts.
  6. Certification — Full TrustVerdict: Mystery Shopper + GuardScan + identity verification. Tiers: Unverified → Verified (50+) → Gold (70+) → Diamond (90+).

Commands

guard scan [path]

Scan a directory for security issues. Checks 102 patterns including:

  • Hardcoded API keys and secrets
  • Prompt injection vulnerabilities
  • PCI DSS violations
  • Sensitive file access (.ssh, .env, private keys)
  • Unsigned or unverified dependencies
guard scan .
guard scan ~/.openclaw/skills/
guard scan /path/to/agent/code

Output: Risk score 0-100, categorized findings, remediation steps.

guard shopper <agent_name>

Run 10 adversarial probes against an agent:

ProbeWhat It Tests
basic_taskCan it follow instructions?
malformed_inputDoes it handle garbage safely?
ethical_boundaryWill it refuse fraud requests?
timeout_testDoes it respond in time?
data_handlingDoes it leak PII?
capability_verifyCan it do what it claims?
idempotencyWill it double-charge?
concurrencyDoes it handle parallel requests?
statefulnessDoes it maintain context?
resource_consumptionIs it efficient?
guard shopper MyCoolAgent
guard shopper MyCoolAgent --endpoint https://myagent.com/api/probe

Output: Score 0-100, pass/fail per probe, trust tier.

guard score

Calculate GuardScore for a merchant:

guard score --chargeback-ratio 0.8 --vertical crypto --volume 50000

Output: Score 0-100, health band (SAFE/WARNING/ELEVATED/CRITICAL), factor breakdown, action items.

guard coach <vertical> "<question>"

Ask one of 14 compliance coaches:

Metadata

Stars1401
Views0
Updated2026-02-24
View Author Profile
AI Skill Finder

Not sure this is the right skill?

Describe what you want to build — we'll match you to the best skill from 16,000+ options.

Find the right skill
Add to Configuration

Paste this into your clawhub.json to enable this plugin.

{
  "plugins": {
    "official-merchantguardops-merchantguard": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.