Back to skill

Security audit

xinjianxue-skill-job_change-cn

Security checks for vulnerabilities and agentic risk

Overview

This skill is transparent about using an external paid service, but it asks users to bind an account persistently and sends sensitive personal profile details for career guidance.

Review carefully before installing. Only use it if you are comfortable linking your 心鉴学 account to the agent, storing service credentials in a secure secret store, spending service credits, and sending birth, sex, and location details to the vendor. Avoid submitting other people’s data unless they have directly agreed.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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 (3)

other

Warning
Location
SKILL.md:199
Finding

Disproportionate Collection and External Transmission of Personal Data

Content
View full analysis
⚠️ **开始前需知**:本服务接口调用时会把分析对象的出生年月日(及时分)、性别、出生地、当前所在地等个人信息发送至心鉴学接口(`chinaapi.xinjianxue.com`)处理,但仅做本次服务分析,而不会进行存储;本 AI 以你的账号身份调用,按次消耗体验积分。**请确认已知悉并同意后再开始。** ``` ```markdown ## §5. 报告入参规范 - 分两种输入档,**积分消耗都按 b(基础版)计,a 不收取额外积分**: - **a(完整)**:分析对象出生年月日时分 + 性别 + 出生地(国家 + 省/区/市)+ 当前所在地 - **b(基础)**:只有分析对象出生年月日 + 性别 + 当前所在地 - **两档都必须提供**:性别、出生年月日(若用户知道出生时分则连时分一起给,自动升 a 档)、分析对象当前所在地(国家 + 省/区/市) - **a 档另需**:出生具体时 + 分 + 出生地(国家 + 省/区/市) - **多人分析**:传 `people` 数组,每人 `{name, birth, gender, place, country}`。 - ⚠️ 用户未给全时,AI 需先询问补齐,不要缺参硬调。 请求体示例: ```json { "license": "XJX-AI-xxxxxxxxxxxxxxxx", "api_key": "", "birth": "1990-01-01 08:30", "gender": "女", "country": "中国", "place": "浙江省杭州市", "current_country": "中国", "current_place": "浙江省杭州市" } ``` ``` ### Technical Analysis The skill requires the agent to collect a user's sex, precise birth date, potentially precise birth time, birthplace, and current location and transmit those attributes to `chinaapi.xinjianxue.com`. These data points can jointly support detailed profiling and possible re-identification. The stated purpose is job-change guidance, but the skill does not explain why every mandatory demographic and location field is necessary for that purpose. It also offers no local or data-minimized alternative. Although the document instructs the agent to obtain consent and claims that the service does not retain reports, the repository contains no server-side implementation, retention policy, deletion mechanism, or technical evidence through which that claim can be verified. The support for arrays of people further increases the exposure scope. The activation policy requires consent from an analyzed person, but the API schema itself does not provide a mechanism for verifying that consent. ### Attack Pa ...[truncated 1267 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:125
Finding

Persistent External Account Authorization Exceeds Least-Privilege Requirements

Content
View full analysis
「请搜索并打开 **微信小程序 "心鉴学" → 我的权益 → AI 助手授权码**,把显示的 8 位授权码发给我,我需要用它完成账号绑定。 > 提醒:这串码是**敏感的一次性凭证**,绑定后立即作废 —— 请确认你是在跟心鉴学官方 AI 助手对话(对接域名 `chinaapi.xinjianxue.com`),并且只发给我这一个 AI;不要发到群里、不要截图外传、不要给其他 AI。」 `POST /api/xinjianxue/ai/license/bind`,body: ```json { "license": "XJX-AI-xxxxxxxxxxxxxxxx", "auth_code": "<用户给的 8 位码>" } ``` 成功返回 `api_key`(业务凭证)。 > ⚠️ **AI授权码是敏感的一次性凭证**(8 位、大小写敏感),绑定后即作废,**仅限本次绑定用途**。若用户曾把授权码发给过其他 AI 且已被使用,本 AI 将无法用同一账号绑定(该账号已绑定其他 AI,需到小程序联系客服处理)。不要替用户猜测或反复尝试。 ### 第 3 步:把凭证收好,之后每次调用带上 把 `license_key` + `api_key` 存进**专用密钥存储**(环境变量,或系统密钥链 / 凭据管理器;与用户协商确定具体位置),每次调用时读取。**不得**写进对话、笔记、配置文件或模型长期记忆: ```json { "license_key": "XJX-AI-xxxxxxxxxxxxxxxx", "api_key": "" } ``` 存好后**再调一次 verify** 复核 `pass: true`,然后向用户确认:「已绑定成功,以后我会以你的账号身份使用服务。」 ``` ### Technical Analysis The skill directs the agent to solicit a sensitive one-time authorization code, bind an external user account to an agent-controlled license, obtain an API key, and retain the resulting credential pair for subsequent sessions. The credentials authorize report requests that consume account credits and billing-history requests that expose account activity. This creates durable account authority where a single-use, narrowly scoped authorization would be sufficient for an individual job-change analysis. The document recommends secure credential storage and prohibits displaying credentials in conversation, which reduces accidental disclosure, but it does not eliminate the excessive authorization lifetime or scope. The binding is also described as difficult to reverse or transfer without contacting customer support. That increases the consequences of binding the wrong agent, ...[truncated 1388 chars]
Remediation
View remediation

T01 · Skill Instruction Hijacking

Note
Location
SKILL.md:272
Finding

Mandatory Vendor Promotion Embedded in Error Handling

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep