玄学大师

• Outcome: Review
• Latest audit: Aug 19, 2026 · 2:55 AM UTC
• Version: 0.1.1

Overview

The skill appears to do fortune-telling as advertised, but it may collect exact birth details, location, and photos and create report files without clear upfront privacy and retention controls.

Review this skill before installing if you would not want an agent to ask for exact birth details, location, home-layout information, or face/palm photos. Use it only for entertainment or cultural reference, avoid sharing unnecessary personal images or exact identifiers, and do not rely on its output for medical, legal, financial, or major life decisions. If generating reports, treat the files as sensitive and delete them when no longer needed.

SkillSpector

Vulnerability Patterns

• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (9)

Description-Behavior Mismatch

Medium

• Confidence: 96% confidence
• Finding: The function accepts a `calendar` parameter and the CLI advertises both `solar` and `lunar`, but the implementation always feeds the supplied date into `solar_to_lunar(...)` as if it were a Gregorian date. In this skill context, that creates silent integrity failures: users who provide lunar dates will receive a chart computed from the wrong birth date, which can mislead downstream reports and recommendations without any visible error.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The global trigger list is extremely broad and includes common terms such as 运势、健康运、事业运、财运、星座、起名, which can appear in ordinary conversations. Overbroad activation can cause the wrong skill to engage, leading users to disclose sensitive personal data or receive domain-specific advice they did not intend to request.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: Several module-level trigger words lack boundary constraints and overlap with normal language, increasing the chance of accidental routing into sensitive workflows. In this skill, accidental activation is more dangerous because many flows immediately ask for birth details, photos, or other personal information.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill instructs collection of highly sensitive data including birth date/time, gender, location, and photos for face/palm analysis, but does not present a clear upfront notice about necessity, retention, sharing, or minimization at collection time. This creates privacy risk because users may reveal biometric-like images and detailed personal profile data without informed consent or understanding how the information will be used.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The workflow states that the skill will run local scripts and generate HTML report files, but does not clearly tell the user before these actions occur. Lack of execution/output transparency can undermine user consent and increases risk if generated files contain sensitive personal data or are stored/shared unexpectedly.

Missing User Warnings

Low

• Confidence: 92% confidence
• Finding: The file only includes a generic disclaimer that tarot results are 'for reference' and should be viewed rationally, but it does not explicitly warn users not to rely on the guidance for medical, legal, financial, or other high-stakes decisions. In the context of a fortune-telling skill that is designed to advise on life events, users may over-trust the output and delay professional help or make harmful decisions based on divination-style guidance.

Natural-Language Policy Violations

Low

• Confidence: 91% confidence
• Finding: The document explicitly maps yin and yang to 'male/female' as a general rule without context, qualification, or user choice. In an interactive divination or advisory skill, this can propagate gender-essentialist assumptions into personalized outputs, leading to exclusionary or inappropriate guidance for users whose identity does not fit the binary framing.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# lunardate 用于紫微斗数排盘的精确农历转换, 未安装时自动降级到内置数据表
# lunar_python 用于八字排盘的精确节气计算(年柱/月柱分界), 未安装时降级到近似公式
# 安装命令: pip install lunardate lunar_python
lunardate>=1.2.0
lunar_python>=1.4.0

• Confidence: 92% confidence
• Finding: The dependency is specified with a lower-bound range instead of an exact version, so future installs may resolve to newer package releases with unreviewed changes or newly introduced supply-chain issues. While this file does not itself execute code, unpinned dependencies reduce build reproducibility and increase the risk of pulling a compromised or breaking version during installation.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# lunar_python 用于八字排盘的精确节气计算(年柱/月柱分界), 未安装时降级到近似公式
# 安装命令: pip install lunardate lunar_python
lunardate>=1.2.0
lunar_python>=1.4.0

• Confidence: 92% confidence
• Finding: Using `>=` for this package allows installation of any later version, which can introduce unreproducible environments and expose consumers to upstream malicious releases, dependency confusion, or accidental security regressions. In a skill package, this is mainly a software supply-chain hygiene issue rather than an immediately exploitable flaw in business logic.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.