Back to skill

Security audit

xhs-comments

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because scanners reported exposed reusable account credentials, disabled TLS certificate checks when enabled, and extra collection and payment capabilities beyond its stated purpose; the review is incomplete.

Consider avoiding installation until the credential and TLS concerns are addressed and the extra collection and recharge behavior is clearly disclosed and controlled. Skillspector coverage was partial: it scanned 2 of 8 components, with 6 only partially inspected, so the evidence is incomplete.

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

Error
Location
src/config.js
Finding

Default HTTP transport exposes reusable account credentials

Content
View full analysis

Impact

A captured bearer token can be replayed to impersonate the user to the platform, access account-scoped collection results, and invoke credit-consuming operations.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The Xiaohongshu comment endpoints align with the description, but the client also implements callable collection workflows for other content types and platforms that the description does not disclose. These are substantive additional capabilities rather than authentication or networking support. The file appears to be a shared library, so it does not establish that these additional methods are automatically invoked or exposed by the skill's entry point. Authentication, source metadata, timeouts, and transport configuration are supporting implementation details and are not the basis for this mismatch finding.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

该代码片段的直接行为是请求注销,而非采集或处理小红书评论。尽管声明提到会话制,主动退出登录会改变认证或会话状态,不能仅凭这一表述视为评论采集的常规实现细节。由于未提供 cli.js 的实现,无法确认注销的具体范围或是否成功,但 logout 调用本身与声明的功能不符。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The description discloses point consumption but does not disclose automatic recharge-order creation or an integrated payment workflow. Creating a monetary order is a substantive additional capability, rather than merely an internal detail of collecting comments. The code does not itself complete payment or automatically debit funds; the user must pay separately. The absence of comment-collection logic in this shared module is not independently evidence of a mismatch.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The activation description includes generic mentions of “用户评论” (user comments) and “舆情” (public sentiment) as standalone triggers.

Medium
Category
Trigger Abuse
Confidence
95% confidence
Finding

The description says to activate when these terms are mentioned, independently of its more specific Xiaohongshu collection conditions. These terms also occur in unrelated feedback and sentiment-analysis requests, so they can cause unintended invocation.

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
99% confidence
Finding

When invoked, allowInsecureTls disables TLS certificate validation process-wide, allowing a network attacker to impersonate HTTPS services and intercept bearer tokens or modify API responses. This is conditional rather than a demonstrated default, but its global effect extends beyond the intended backend, and warning messages do not restore transport authentication.

Content

Scanner excerpt · src/client.js (reported line 75)May include surrounding context.

js
/** 忽略 HTTPS 证书校验 (服务端自签/域名不匹配证书时由 insecure 配置开启)。进程级。 */
function allowInsecureTls() {
  process.env.NODE_TLS_REJECT_UNAUTHORIZED = '0';
}

/** 业务错误: code 为后端错误码 (1001/1002/3001/...) */

The Xiaohongshu-comment skill exposes unrelated profile, danmaku, and multi-platform collection capabilities.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest limits the skill to collecting and exporting Xiaohongshu note comments, but the client also implements profile-note collection (L178–L205), Douyin danmaku collection (L207–L233), and comment collection across numerous other platforms (L284–L519). These methods can create and advance remote collection sessions using the client's authentication token, and are accessible through the exported ApiClient (L523–L525); they are not necessary to implement the declared Xiaohongshu-comment workflow. Their presence establishes additional callable capabilities, although this file does not show them being invoked automatically.

Content

No source excerpt is available for this finding.

Insufficient credits automatically trigger creation of a payment order, beyond the manifest's disclosed per-page credit deductions.

Medium
Category
Excessive Agency
Confidence
87% confidence
Finding

The manifest discloses paid comment collection but does not disclose automatically initiating recharge orders. On an insufficient-credit error, handleFatal invokes rechargeFlow without confirmation; that flow selects a payment channel and amount and POSTs a new recharge order (L107–L124), even in noninteractive agent execution. This creates a financial-service resource, although payment itself still requires the user to scan and pay.

Content

No source excerpt is available for this finding.

The registration instructions require verbatim Chinese user-facing output without language opt-in.

Low
Category
Output Handling
Confidence
90% confidence
Finding

The agent must send the registration guidance exactly as written, and the supplied guidance is entirely Chinese. Unlike merely documenting the skill in Chinese, this mandates a specific output language without offering a language choice or justifying a locale restriction.

Content

No source excerpt is available for this finding.