Back to skill

Security audit

Ai Exam 授客AI智能考试

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly does what it claims, but it sends organization credentials, documents, user data, and access tokens to a configurable remote API without enough destination validation or privacy disclosure.

Install only if you trust the configured API endpoint and the organization credentials have limited scope. Set BASE_URL only to the approved HTTPS service, avoid uploading confidential or regulated documents unless authorized, and treat exam results and user/dept lookups as sensitive personal or organizational data. Prefer a reviewed lockfile or pinned dependencies before production use.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
api-client.js:20
Finding

Credentials and Access Tokens Are Transmitted to an Unrestricted Destination Through URL Query Parameters

Content
View full analysis
Remediation
View remediation
` header rather than `access_token` query parameters. 4. **Harden initialization** - Accept a validated base URL through an explicit initialization configuration object if runtime configuration is required. - Update the API client's base URL only after validation. - Fail closed when the base URL or required credentials are absent. 5. **Protect logs and errors** - Redact authorization headers, application secrets, tokens, and sensitive query parameters from application, proxy, and monitoring logs. - Avoid propagating complete request URLs in user-visible error messages. 6. **Rotate potentially exposed credentials** - Rotate the application secret and invalidate active tokens after deploying the fix. - Review server, proxy, and observability logs for previously recorded secrets or tokens. A hardened design should resemble: ```javascript const allowedOrigin = 'https://api.example.invalid'; const parsedBaseUrl = new URL(process.env.BASE_URL); if (parsedBaseUrl.origin !== allowedOrigin) { throw new Error('Unapproved API origin'); } const response = await axios.post( `${allowedOrigin}/service/corp/g ...[truncated 285 chars]

T08 · Insecure Dependencies

Note
Location
package.json:18
Finding

Unpinned Dependencies and Missing Lockfile Make Installations Non-Reproducible

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The documented behavior extends beyond the high-level description by including user and department lookup operations against organizational directory data, yet those capabilities are not clearly disclosed in the declared purpose. This kind of description-behavior mismatch can mislead operators about what data the skill accesses and can result in unintended processing of internal directory information.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The README promotes uploading documents, assigning exams to users/departments, and querying student results, but it does not warn that these actions may transfer personal data, educational records, or confidential document contents to external services. In an AI exam system, this omission can cause operators to process regulated or sensitive data without informed consent, minimization, or governance controls, increasing privacy and compliance risk.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill documentation describes use of environment variables for APP_KEY, APP_SECRET, and CORP_ID, which implies access to sensitive runtime secrets, but it declares no explicit tool scope or permission boundary. In an agent setting, undocumented secret access increases the risk of overprivileged execution and makes it harder for users or reviewers to understand what data the skill can read.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill encourages uploading documents and handling exam, learner, and result data, but it does not warn users that these materials may be transmitted to external APIs or third-party services. This omission can cause sensitive educational, HR, or corporate content and personal data to be sent off-platform without informed consent or governance review.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The code sends sensitive credentials (app_key, app_secret, and corpid) in an HTTP request to obtain an access token, but there is no visible confirmation, logging, or comment warning users that credentials will be transmitted to a remote service. For code files, network transmission of sensitive data should have some form of user disclosure unless it is clearly documented as the skill's stated purpose, which is not evident in this file alone.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The uploadFile method reads a local file from disk and sends it via POST to /ai/uploadFiles. Although the docstring says the method uploads a file, there is no confirmation prompt, logging, or explicit warning that local file contents and associated metadata will be transmitted to an external service.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The createExam method constructs a request containing user_id, user_name, dept_user_id, and dept_user_name and submits it to /exam/exam/create. The code lacks any confirmation, visible notice, or warning that personally identifying operator data is being transmitted.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The assignExam method sends assigned user IDs, department IDs, and operator identity fields to /exam/user/assign. There is no prompt, log, or inline warning to disclose that assignment and identity data will be sent to a remote service.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

This method uploads a document, creates an exam from it, queries user and department records, and assigns the exam via remote service calls, which can affect user data and system state. While there are console logs describing progress, they are operational logs rather than a user disclosure or confirmation about transmitting document contents and assignment data to external services.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

All human-readable strings and comments in the file are in Chinese, including error messages that may surface to users, with no indication that language selection is configurable or optional. This can violate a language/locale policy when a skill forces a specific language without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

User-visible names, comments, and console output consistently force a specific language/locale, which can violate language-choice policy when no opt-in or justification is provided. The file does not indicate that the skill is intentionally restricted to a Chinese-speaking or region-specific environment.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
96% confidence
Finding

The dependency uses a caret range instead of an exact pinned version, which allows different installs over time to resolve to different package releases. This weakens build reproducibility and can expose the skill to newly introduced vulnerable or malicious upstream versions without any code change in the repository.

Content

Scanner excerpt · package.json (reported line 19)May include surrounding context.

json
"author": "",
  "license": "MIT",
  "dependencies": {
    "axios": "^1.6.0",
    "dotenv": "^17.3.1",
    "form-data": "^4.0.0"
  }

Unverifiable Dependency: axios has 16 known advisory(ies) (CVE-2026-44494 (axios Vulnerable to Full Man-in-the-Middle via Prototype Pollution Gadget in `co); CVE-2026-44495 (axios Vulnerable to Credential Theft and Response Hijacking via Prototype Pollut); CVE-2025-62718 (Axios has a NO_PROXY Hostname Normalization Bypass that Leads to SSRF) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
89% confidence
Finding

Axios has multiple known advisories, and because the manifest does not pin the exact installed version, it is impossible to verify from this file whether the deployed package is affected. This uncertainty is risky for a network-facing skill, since axios is commonly used for outbound HTTP requests and some listed issues include SSRF and traffic manipulation scenarios.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

Using a non-pinned dotenv version permits automatic uptake of later compatible releases, reducing supply-chain control and reproducibility. If an upstream release is compromised or introduces a security flaw, deployments may silently consume it.

Content

Scanner excerpt · package.json (reported line 20)May include surrounding context.

json
"license": "MIT",
  "dependencies": {
    "axios": "^1.6.0",
    "dotenv": "^17.3.1",
    "form-data": "^4.0.0"
  }
}

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
96% confidence
Finding

The form-data dependency is specified with a caret range, so installations are not deterministic and may drift to later releases. In a skill that likely handles document uploads and exam content generation, weakened dependency integrity increases supply-chain risk around sensitive data processing.

Content

Scanner excerpt · package.json (reported line 21)May include surrounding context.

json
"dependencies": {
    "axios": "^1.6.0",
    "dotenv": "^17.3.1",
    "form-data": "^4.0.0"
  }
}

Unverifiable Dependency: form-data has 2 known advisory(ies) (CVE-2025-7783 (form-data uses unsafe random function in form-data for choosing boundary); CVE-2026-12143 (form-data: CRLF injection in form-data via unescaped multipart field names and f)), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
87% confidence
Finding

Form-data has known advisories, and the unpinned manifest prevents verification of whether the installed version includes the fixes. For a system that likely uploads or forwards documents, unresolved multipart handling issues could affect request integrity or enable injection-style abuse depending on how user-controlled fields are used.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

This JavaScript code sends department-name query data to a remote API, and the file only contains developer-oriented comments about functionality. For code-file SQP-2, network calls that transmit user or system data should have some visible disclosure such as a prompt, log, or explicit warning unless that warning is provided elsewhere.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The function posts user-name search terms to a backend endpoint, which is a network transmission of potentially identifying user data. The file has no visible confirmation, logging, or warning text informing users of this transmission, so it meets the code-file criteria for a missing disclosure finding.

Content

No source excerpt is available for this finding.