Back to skill

Security audit

shixinchao

Security checks for vulnerabilities and agentic risk

Overview

This skill is a broad auto-triggering Q&A wrapper that sends user questions to an external API, but its name, description, consent language, and credential handling are too unclear for the access it requests.

Review this carefully before installing. Treat anything you ask through this skill as being sent to a third-party service, do not include secrets, personal data, proprietary text, or sensitive files, and verify the API provider and credential setup first because the script does not appear to load the documented API_KEY environment variable.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (10)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared name and description are effectively non-descriptive ('测试-史新超'), while the skill actually sends user questions to a remote chat API using API-key authentication and returns remote-generated content. This mismatch conceals material behavior from users and reviewers, increasing the risk of unintended data disclosure and unauthorized external processing.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The trigger condition states the skill auto-triggers whenever a user asks a question, which is broad enough to match nearly all conversations. Because the skill forwards input to an external API, this broad activation scope can cause inadvertent transmission of sensitive or irrelevant user content without deliberate invocation.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill invokes a Python script that performs external network access, but the manifest does not declare any tool scope such as permissions or allowed-tools. This weakens platform governance and user trust because network-capable behavior is present without explicit capability declaration, making data egress easier to hide or misconfigure.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill documentation notes API key usage and an external endpoint internally, but it does not provide a clear user-facing warning that user questions are transmitted to a third-party service for processing. This reduces informed consent and may expose sensitive prompts to an external provider without the user's awareness.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file documents uploading user-provided files to a remote service and authenticating with a bearer token, but it does not include any user-facing warning about sensitive data exposure, handling of uploaded content, or the need to protect credentials. Under the markdown-specific SQP-2 criteria, documentation should warn when behavior can affect user data, privacy, or system integrity.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The code is designed to use bearer-token authentication and external network access, but the surrounding description does not justify these capabilities or explain their necessity. Undocumented sensitive capabilities increase the risk of credential misuse and covert data exfiltration under the guise of a harmless test skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script’s stated purpose is only a generic test label, but the implementation transmits user-supplied content to an external async chat service. This mismatch reduces user awareness and informed consent, making unintended data disclosure more likely when users provide sensitive prompts.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

This code performs an outbound HTTP POST to a remote service carrying user input and request variables. External transmission is risky in this skill context because the skill is ambiguously described, so users are not adequately informed that their data leaves the local environment.

Content

Scanner excerpt · scripts/program.py (reported line 73)May include surrounding context.

python
# 4. 发起请求
    try:
        response = requests.post(
            url,
            headers=headers,
            json=payload,

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The script sends the user-provided question directly to a third-party API without any user warning, consent prompt, or sensitivity check. If a user enters secrets, personal data, or proprietary content, the program will silently transmit that data off-host.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The entire skill document is presented in Chinese and does not offer a language choice, multilingual alternative, or justification for a locale-specific restriction. Per SQP-3, forcing a specific language without user opt-in can constitute a natural-language policy violation.

Content

No source excerpt is available for this finding.