Back to skill

Security audit

Xuanself

Security checks for vulnerabilities and agentic risk

Overview

The skill is not clearly malicious, but it needs Review because it asks for broad web, command, and file authority while also encouraging plaintext API keys and loading exporter code from a sibling directory.

Review this before installing. Use an isolated Python environment, pin dependencies, do not store real API keys in data/data_sources.json, and only run it after checking any installed report-gama sibling directory. Treat generated searches as third-party data sharing and confirm report output paths before use.

Vulnerability Patterns
  • Tool Hijacking and SpoofingModifies or replaces tools so legitimate-looking calls execute attacker logic
  • 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
Findings (3)

T07 · Tool Hijacking and Spoofing

Error
Location
scripts/run_report.py:9
Finding

External Module Precedence Allows Local Tool Hijacking

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
README.md:20
Finding

Unpinned Third-Party Dependency Installation Creates Supply-Chain Exposure

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
data/data_sources.json:9
Finding

Plaintext API Credentials Are Stored in a Packaged Project File

Content
View full analysis
Remediation
View remediation
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 Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (12)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description presents a full research and monitoring workflow for the Russian medical-device market, including multiple external data-gathering capabilities and generation of a delivery-ready Word report. However, the supplied code chunk only implements a local export utility: it validates command-line arguments, checks for a Markdown file, imports a WordExporter from another path, and converts Markdown to DOCX. This supports only the final document-conversion aspect and does not implement the core declared functionality. Therefore the code's actual behavior is materially narrower and different from the stated purpose.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

描述将该技能定位为完整的俄罗斯医疗器械市场深度调研工作流,核心应包括数据检索、监测、整合与分析。实际代码却只是一个 Word 导出模块,负责把 Markdown 内容转换为排版较专业的 DOCX 文档。它没有网络访问、外部 API 调用、数据抓取、市场分析、患者数据处理或政府动态监控等行为。虽然“输出 Word 报告”与声明中的最终交付形式部分相关,但这只是整体流程中的一个配套导出环节,不能代表所宣称的主要能力,因此属于明显的描述与行为不一致。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest sets "language": "zh-CN", which indicates a fixed Chinese locale. Under the policy, forcing a specific language without user opt-in or a clearly documented justification is a natural-language policy violation, and this manifest does not provide either.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The trigger phrases are very broad and imply the skill may auto-run on generic requests about Russia medical-device reports without clear user-consent or scope boundaries. In an agent environment, ambiguous activation can cause unintended web queries, API usage, and report generation based on casual prompts, increasing the chance of unapproved external calls or file creation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README advertises automatic real-time collection from external sources and Word export but does not warn that user requests or report content may be transmitted to third-party services or written to local storage. This can mislead users about privacy and side effects, especially in enterprise or regulated environments where outbound data flow and local artifact creation require explicit notice and approval.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill advertises use of WebFetch, WebSearch, ExecuteCommand, file writing, and third-party services such as SerpAPI/TGStat, but the user-facing description does not clearly disclose that prompts, queries, or generated report content may be transmitted externally or written to local files. In a market-research workflow that may contain proprietary strategy, customer names, pricing assumptions, or regulated health-related context, this omission can cause unintentional data exposure and unsafe operator assumptions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The description states this is a core data-source index for Russian medical-device market research, and the file subsequently hard-codes Russian-language search keywords and mostly RU-language sources. Because there is no natural-language indication that language/locale is optional or user-selected, this can be read as enforcing a specific locale without opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The default title, footer text, and multiple fixed strings are written in Chinese, and the module also fixes Chinese fonts as defaults. This creates a language/locale constraint in natural-language output without any opt-in or user-selectable locale behavior, which matches the policy-violation category.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

该技能文档全文使用中文,未见提供其他语言选项、用户可选择语言的说明,或对必须使用中文的业务理由进行说明。根据规则,强制特定语言而无用户选择属于自然语言层面的语言/区域政策问题。

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger section lists several phrases and then appends '等', implying additional unspecified triggers. For a manifest/markdown skill description, this leaves the invocation boundary ambiguous and could cause unintended activation beyond the explicitly listed phrases.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The module docstring and CLI usage text are written only in Chinese, and later user-facing status/error messages follow the same pattern. This imposes a specific language on users without opt-in or an explicit justification, which matches the locale-policy concern for natural-language content.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The manifest describes a specialized medical-device Russia market research workflow that outputs professional Word reports. This file also acts as a generic local Markdown-to-DOCX utility via main and quick_export, allowing arbitrary user-supplied file conversion outside that domain-specific workflow.

Content

No source excerpt is available for this finding.