Back to skill

Security audit

微信生态集成 · WeChat Integration

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious because it may allow a production payment test that could charge real money without requiring explicit approval; the review found no reported malicious behavior, but coverage is partial.

Review the payment-testing guidance before installing. In particular, confirm that any production test explains possible real charges, fees, and fulfillment effects and requires your approval; prefer sandbox testing. The scan was partial because three local references could not be resolved, so the evidence does not cover those references.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% 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.

Content

Scanner excerpt · SKILL.md (reported line 60)May include surrounding context.

md
1. Treat Integration Center creation as a console-first workflow unless a public Manager SDK or Cloud API contract is confirmed in official docs.
2. Use official `index.md` docs for console UI steps and credential fields; do not copy stale console screenshots or invent field names.
3. Never ask the user to paste secrets into chat. Tell them to configure merchant and official-account credentials in the CloudBase console Integration Center form.
4. Do not assume generated function names are fixed. `pay-common` and `offiaccount-common` are examples; ask for or inspect the actual function name before writing calls.
5. Treat frontend payment success as UI feedback only. The authoritative payment state must come from server-side query results or payment callbacks.
6. When extending generated functions, preserve credential environment variables and generated callback verification/decryption logic. Add business logic around order checks, persistence, idempotency, and fulfillment.

The workflow permits a production payment test without warning that it can charge real money.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The workflow includes a low-value production payment test without explicitly warning that it charges real money or requiring authorization before initiating the transaction. In an agent-operated payment integration, this omission can lead to unintended charges, fees, or fulfillment effects; limiting the transaction value reduces but does not eliminate that risk.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/virtual-payment.md (reported line 116)May include surrounding context.

md
### Query order (查单兜底)

`POST https://api.weixin.qq.com/xpay/query_order?access_token=%2A%2A%2A&pay_sig=PAY_SIG%60

`pay_sig` 的 uri 是 `/xpay/query_order`(不带 query),消息体是实际发出的原始 JSON。本接口不支持云调用。

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/virtual-payment.md (reported line 156)May include surrounding context.

md
### refund_order(只表示任务已受理)

`POST https://api.weixin.qq.com/xpay/refund_order?access_token=%2A%2A%2A&pay_sig=PAY_SIG%60

`pay_sig` 的 uri 是 `/xpay/refund_order`。调用 URL 只带 `access_token` 与 `pay_sig`,不要另造用户态签名 query。本接口不支持云调用。