ClawHub
Back to skill

Security audit

Halocard Virtual Credit cards

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed payment helper for creating single-use virtual cards, with explicit confirmation and spending-limit safeguards before purchases.

Install only if you trust Halocard and are comfortable giving an agent access to create virtual cards for checkout. Keep HALOCARD_TOKEN protected, confirm merchant and total carefully, and avoid using the skill where returned card details could be logged or exposed.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
3. **Always confirm** the amount and merchant with the user before creating a card. No exceptions.
4. Create the card **only when ready to fill** the payment form — cards are single-use.
5. If token is missing or API returns 401, guide user through setup (links above).
6. If payment fails, report the error to the user. Do not retry with a new card without asking.

## Example (curl)
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.

External Transmission

Medium
Category
Data Exfiltration
Content
## Example (curl)

```bash
curl -X POST https://agent.halocard.co/api/v1/payments \
  -H "Authorization: Bearer $HALOCARD_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"amount": 2450, "description": "Uber Eats order - 2x chicken shawarma"}'
Confidence
96% confidence
Finding
The skill instructs the agent to transmit a bearer token and purchase metadata to an external payment API and to retrieve full card details (PAN/CVC/expiry) for use in a checkout flow. In a payment skill this behavior is expected, but it is still security-sensitive because it enables external fund movement and exposes highly sensitive payment credentials to the agent runtime and browser context if mishandled.

Static analysis

No suspicious patterns detected.