Back to skill

Security audit

1688 Distributor

Security checks for vulnerabilities and agentic risk

Overview

This skill openly automates 1688 product distribution, but it can bulk-submit shop listings and copy logs with broad triggers and weak confirmation controls.

Install only if you are comfortable with an agent operating an authenticated 1688 merchant session. Before use, require explicit confirmation for the target shop, product count, selected filters, and final distribution action, and avoid running it from vague product-research requests. Review any copied logs for sensitive shop or product details before sharing them elsewhere.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:92
Finding
Unsafe DOM Targeting Can Trigger Unintended Product Distribution## Vulnerability Details **File Location**: `SKILL.md`, lines 92–123 **Vulnerability Type**: Ambiguous DOM selection for transactional actions **Risk Level**: Medium ### Vulnerable Code ```javascript // Find and click the "Distribute Now" button const distributeBtn = [...document.querySelectorAll('button, div[role="button"]')].find(el => el.textContent && el.textContent.includes('立即铺货') && el.offsetParent !== null ); if(distributeBtn) { distributeBtn.scrollIntoView({ behavior: 'smooth', block: 'center' }); distributeBtn.click(); } else { console.error('未找到"立即铺货"按钮'); } ``` ```javascript // Find and click the target shop (replace shop name) const targetShop = "kyeshop小店"; // Obtain from user input const shopElements = [...document.querySelectorAll('*')].filter(el => el.textContent && el.textContent.includes(targetShop) && el.offsetParent !== null ); if(shopElements.length > 0) { shopElements[shopElements.length - 1].click(); } // Find and click the confirmation button const confirmBtn = [...document.querySelectorAll('*')].find(el => el.textContent && el.textContent.includes('立即铺货') && el.offsetParent !== null ); if(confirmBtn) confirmBtn.click(); ``` ### Technical Analysis The automation identifies consequential controls using broad, partial text matching. In particular, `querySelectorAll('*')` examines every element in the document, including noninteractive containers, nested descendants, duplicated text, and marketplace-controlled content. It then chooses the last shop-name match and the first visible element containing the distribution label. Visibility through `offsetParent !== null` does not establish that an element is the intended control. The code also does not require an exact normalized text match, a unique match, an expected dialog ancestor, or a verified button role. Before confirmation, it does not verify the selected shop, selected product count, or final transaction summary. This creates an exploitable confused-control c ...[truncated 1575 chars]
Remediation
## Remediation Suggestions 1. Replace global `querySelectorAll('*')` searches with stable selectors scoped to the expected modal, form, or table. 2. Restrict candidates to appropriate interactive elements, such as `button`, elements with a verified button role, or documented application controls. 3. Compare exact normalized text rather than using unrestricted substring matching. 4. Require exactly one valid match. Abort and report an error when no match or multiple matches are found. 5. Identify shops using stable application identifiers rather than display-name text where possible. 6. Verify after selection that the displayed shop identifier or exact shop name equals the user-requested destination. 7. Before the final submission, validate the selected product count, destination shop, and operation type against an expected summary. 8. Require explicit user confirmation immediately before the irreversible distribution action. 9. After submission, inspect a structured success response or verified status element rather than inferring success merely from a click. 10. Add safe failure behavior: do not proceed to confirmation if any earlier selection or verification step is ambiguous.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Missing User Warnings

High
Confidence
96% confidence
Finding
This section automates selecting all items and submitting '立即铺货' to a target shop without a clear user warning or mandatory confirmation immediately before execution. In context, the action affects a real storefront and can create unwanted product listings, operational burden, policy violations, or financial/reputational harm at scale.

Vague Triggers

Medium
Confidence
96% confidence
Finding
The trigger conditions are broad enough that casual mentions of 1688-related shopping or providing a shop name and criteria could invoke a workflow that performs high-impact commercial actions. Because the skill can proceed from search to bulk distribution, accidental activation could cause unintended listing/distribution operations in a merchant account.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The example trigger phrase '去1688找点货' is vague and can reasonably mean browsing or research, yet this skill is designed to continue into selection and distribution actions. That mismatch increases the chance that a user request for exploration is interpreted as authorization to perform bulk shop modifications.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The workflow copies distribution logs without warning that logs may contain shop identifiers, product details, operational status, or other sensitive business data. Automatically copying and exposing that content can leak private merchant information into chat history, downstream tools, or the system clipboard.

Static analysis

No suspicious patterns detected.