Back to skill

Security audit

Ntriq X402 Invoice Extract

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a paid invoice-extraction API, but its privacy claim conflicts with sending sensitive invoice images or URLs to a third-party endpoint.

Review this carefully before installing or using it with real invoices. Do not submit confidential, regulated, customer, tax, banking, or presigned-URL documents unless you are allowed to share them with the service operator and have acceptable privacy, retention, and deletion terms.

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
Findings (1)

other

Warning
Location
SKILL.md:14
Finding

Misleading Privacy Claim May Cause Sensitive Invoice Disclosure

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 14-28
Vulnerability Type: Sensitive Data Disclosure and Misleading Privacy Claim
Risk Level: Medium

Vulnerable Code Snippet

markdown
Turn invoice or receipt images into structured JSON — vendor, customer, line items, tax, totals, payment terms. Local AI vision, no cloud upload, no API key. Pay $0.03 USDC per call via x402 (Base mainnet).

## How to Call

```bash
POST https://x402.ntriq.co.kr/invoice-extract
Content-Type: application/json
X-PAYMENT: <x402-payment-header>

{
  "image_url": "https://example.com/invoice.jpg"
}

Parameters

ParameterTypeRequiredDescription
image_urlstring✅ (or base64)URL of invoice/receipt image
image_base64string✅ (or url)Base64-encoded invoice image
languagestring❌Output language (default: en)
text

### Technical Analysis

The statement “no cloud upload” conflicts with the documented request flow. The Skill directs an agent to contact the external domain `x402.ntriq.co.kr` and provide either:

- The invoice itself as Base64-encoded request data; or
- A URL from which the external service can retrieve the invoice.

In either mode, invoice information is disclosed to, or made accessible by, a third party. Invoices can contain personal and confidential financial data, including names, postal addresses, tax identifiers, customer relationships, transaction details, prices, and payment terms.

The documentation does not provide a privacy policy, retention period, deletion process, access-control guarantees, processing jurisdiction, redaction workflow, or requirement to obtain informed user approval before transmission. The misleading local-processing claim can therefore cause users or agents to submit documents under an incorrect understanding of how their data is handled.

### Attack Pa
...[truncated 1094 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove the “no cloud upload” claim unless all processing occurs entirely on the user's device.
  2. Clearly state that invoice content or its retrieval URL is transmitted to x402.ntriq.co.kr.
  3. Require explicit, informed confirmation before sending a document to the third-party service.
  4. Publish and link applicable privacy, retention, deletion, subprocesser, security, and processing-jurisdiction policies.
  5. Provide a redaction workflow for tax identifiers, account details, addresses, and other unnecessary sensitive fields.
  6. Warn users not to submit private, presigned, or credential-bearing URLs. Prefer short-lived, narrowly scoped URLs where URL submission is unavoidable.
  7. Validate allowed URL schemes and implement server-side protections against unauthorized internal-resource retrieval.
  8. Offer genuinely local processing for confidential documents or clearly distinguish remote processing from any local AI components.
  9. Define transport-security, storage-encryption, access-control, logging, and deletion requirements for uploaded documents.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly accepts invoice images via URL or base64, and invoices commonly contain sensitive financial, business, and personal information such as addresses, tax IDs, account details, and purchase history. Although the text claims 'no cloud upload,' it still directs users to send data to a remote HTTPS endpoint and does not disclose data handling, retention, logging, or privacy considerations, which can mislead users into sharing sensitive documents without informed consent.

Content

No source excerpt is available for this finding.