赛博命理师

• Outcome: Review
• Latest audit: Sep 16, 2026 · 8:49 AM UTC
• Version: 2.1.0

Overview

This fortune-telling skill has no executable install behavior, but it asks for sensitive personal and image data and directs unsupported health, lifespan, and character judgments from photos without enough safeguards.

Install only if you are comfortable with an entertainment skill that may request birth information, gender, and face or palm photos. Do not use its outputs for medical, hiring, relationship, financial, legal, or other important decisions, and avoid uploading photos of other people without consent.

A.I.G

Vulnerability Patterns

• Insecure Skill Coding Practices: Finds exploitable flaws such as hardcoded secrets or command injection
• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally

Findings (2)

T09 · Insecure Skill Coding Practices

Warning

• Location: references/palm-flow.md:85

Finding

Unsupported Medical and Longevity Inferences from Biometric Images

Content

View full analysis

Vulnerability Details

File Location: references/palm-flow.md:85-101, references/palm-flow.md:303-319, references/face-flow.md:372-393, and references/cross-validation.md:305-335
Vulnerability Type: Unsupported health diagnosis and longevity prediction from facial and palm characteristics
Risk Level: Medium

Evidence

references/palm-flow.md:85-101 contains the following operational rules, translated into English:

text

| Dimension | Evaluation standard | Fortune rating | Explanation |
| Length | Begins between the index finger and thumb and ends at the wrist | A long line indicates longevity | Length represents lifespan |
| Depth | Degree of line depth | A deep line indicates abundant energy | Depth represents vitality |
| Curvature | Size of the arc | A large arc indicates abundant energy | Curvature represents energy |
| Form | Continuous, interrupted, or chained | A continuous line indicates stability | Form represents health |
| Symbol | Island, break, or fork | An island represents disease | Symbols represent problems |

| Form | Characteristic | Determination |
| Long and deep | Extends beyond the middle-finger centerline and is deeply marked | Favorable: longevity and abundant energy |
| Short and deep | Does not reach the middle-finger centerline but is deeply marked | Moderate: abundant energy but limited lifespan |
| Long and shallow | Long but shallow | Moderate: long life but average energy |
| Short and shallow | Short and shallow | Unfavorable: health requires attention |
| Chained | Line resembles a chain | Unfavorable: relatively weak constitution |
| Island | Contains an island-shaped marking | Unfavorable: health problem |
| Break | Contains an obvious break | Unfavorable: major health event |
| Fork | End of the line is forked | Moderate: declining energy in later life |

references/palm-flow.md:303-319 contains the following health mappings, translated into En ...[truncated 5088 chars]

Remediation

View remediation

Remediation Suggestions

1. Remove every rule that maps facial or palm characteristics to disease, organ function, surgery, major health events, or lifespan.
2. Prohibit diagnosis, prognosis, lifespan estimation, and treatment recommendations in SKILL.md and all workflow documents.
3. Restrict image analysis to neutral, directly observable characteristics and clearly identified entertainment content.
4. Do not describe agreement among Bazi, face reading, and palm reading as medical cross-validation.
5. Add an explicit rule that visible color changes can be caused by lighting, camera processing, skin tone, temperature, and other non-medical factors.
6. When a user reports symptoms or asks for medical advice, direct the user to a qualified healthcare professional rather than applying fortune-telling rules.
7. Add a health-specific disclaimer explaining that the Skill cannot diagnose conditions or estimate lifespan.
8. Add test cases ensuring that prompts involving anemia, organ disease, surgery, serious illness, or lifespan do not produce conclusions based on images or fortune-telling systems.

other

Warning

• Location: references/face-flow.md:43

Finding

Harmful Personality, Morality, and Criminality Profiling from Facial Appearance

Content

View full analysis

Vulnerability Details

File Location: references/face-flow.md:43-55, references/face-flow.md:118, and references/face-flow.md:168-170
Vulnerability Type: Unsafe biometric profiling
Risk Level: Medium

Evidence

references/face-flow.md:43-55 contains the following profiling rules, translated into English:

text

Evaluation points:
1. Eye clarity: clear eyes indicate nobility; cloudy eyes indicate inferiority.
2. Gaze concentration: a concentrated gaze indicates longevity; a scattered gaze indicates early death.
3. Gaze depth: a concealed gaze indicates depth; an exposed gaze indicates shallowness.
4. Gaze strength: a spirited gaze indicates nobility; a spiritless gaze indicates inferiority.

| Type | Evaluation standard | Characteristic description |
| Clear spirit | Clear and bright eyes | Intelligent, distinguished, and kind-hearted |
| Clouded spirit | Cloudy and dull eyes | Unintelligent, poor, and morally unkind |
| Concentrated spirit | Stable and focused gaze | Steady, long-lived, and thoughtful |
| Scattered spirit | Wandering and unfocused gaze | Drifting life, early death, and mental unrest |
| Concealed spirit | Deep and restrained | Strategic and extremely calculating |
| Exposed spirit | Urgent and expressive | Shallow, impulsive, and straightforward |

references/face-flow.md:118 contains the following eye-shape classification, translated into English:

text

| Snake eyes | Narrow eyes described as poisonous, with reddish pupils | Cruel, malicious, and treacherous |

references/face-flow.md:168-170 contains the following ear-shape classifications, translated into English:

text

| Mouse ears | Small and pointed ears | Petty person associated with theft |
| Pig ears | Large ears without a defined rim | Unintelligent and without good fortune |
| Donkey ears | Large and drooping ears | A lifetime of hardship |

Technical Analysis

The workf ...[truncated 1884 chars]

Remediation

View remediation

Remediation Suggestions

1. Remove all mappings from facial features to intelligence, morality, criminality, maliciousness, social class, poverty, or lifespan.
2. Restrict facial analysis to neutral descriptions of visible characteristics that do not assert sensitive or unobservable traits.
3. Require confirmation that the depicted person consented to the analysis, especially when the image appears to show a third party.
4. Prohibit using the Skill's output for consequential decisions involving employment, housing, education, insurance, credit, healthcare, or law enforcement.
5. Replace derogatory labels with non-judgmental language and clearly state that appearance cannot establish character or intent.
6. Add adversarial tests using requests to determine whether a person is dishonest, dangerous, unintelligent, or likely to commit crimes. Such requests should be declined.
7. Review the remaining classical references for equivalent appearance-based profiling and prevent quoted historical claims from being converted into factual conclusions.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• 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

Findings (21)

Missing User Warnings

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The skill requires full birth date, birth time, birth city, gender, and optionally face and hand photos, but provides no up-front privacy warning, minimization guidance, retention notice, or explicit consent flow. This combines sensitive personal attributes with biometric imagery, creating substantial privacy, profiling, and misuse risk if the data is logged, reused, or exposed.

Content

No source excerpt is available for this finding.

Missing User Warnings

High

• Category: Not specified by scanner
• Confidence: 96% confidence

Finding

The file operationalizes physiognomy by directing analysis of face photos to infer personality, morality, fate, health, and social value from appearance. That is dangerous because it encourages ungrounded sensitive inferences from biometric data and can enable discriminatory profiling, reputational harm, or harmful decision-making without any warning or limitation.

Content

No source excerpt is available for this finding.

Missing User Warnings

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The output template tells the model to produce definitive judgments such as '贵相/贱相,' age-based life outcomes, health implications, and personal strengths or defects from a face alone. In the context of an agent skill, this makes the capability more dangerous because it standardizes and scales sensitive appearance-based profiling while omitting warnings about privacy, bias, or misuse in employment, relationships, or other high-impact contexts.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

The trigger list is very broad and includes common terms like '星座'、'运势' and '算一卦', which increases the chance of accidental activation outside the user's intended scope. In a skill that solicits birth data and optionally face/hand photos, overly broad invocation boundaries can cause unnecessary collection of sensitive data and misleading engagement in contexts where the user did not explicitly opt into this analysis mode.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The activation logic explicitly allows vague requests like '帮我算一卦' or '看看我的命', which makes the invocation boundary ambiguous and encourages the agent to infer consent too loosely. Because the skill then proceeds into structured collection of sensitive birth and image data, unclear activation criteria materially raise privacy and safety risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 92% confidence

Finding

This markdown file presents all operational instructions, examples, and output templates only in Chinese, which effectively forces a specific language for users and downstream agents. The stated policy flags language or locale constraints unless the skill offers user opt-in or clearly documents a justified region-specific limitation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 71% confidence

Finding

All natural-language instructions and required output structure are fixed in Chinese, and the file does not indicate that users may choose another language or that the Chinese-only constraint is intentional for a specific region or audience. The policy rule flags language/locale constraints when they are imposed without opt-in or clear justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 93% confidence

Finding

The document instructs the system to produce health assessments and even longevity expectations from physiognomy/palmistry/astrology without any disclaimer about inaccuracy or lack of medical basis. In a consumer-facing skill, this can mislead users into treating pseudoscientific output as health guidance, delaying appropriate medical care or causing unnecessary anxiety.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 91% confidence

Finding

The file is entirely written as a mandatory execution workflow in Chinese and the required report template is also fixed in Chinese, with no indication that the user may choose another language. Under the stated policy, forcing a specific language without opt-in is a natural-language policy violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

These instructions explicitly map palm lines to lifespan, disease, major health changes, marriage outcomes, and children. This is dangerous because it directs the system to infer highly sensitive personal information and medical-style conclusions from an image without evidence, which can cause emotional harm or harmful real-world decisions.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

This section instructs the model to produce health, vitality, and life-outcome judgments from palm appearance, including warnings about illness and luck. That creates ungrounded sensitive inferences from biometric imagery and can mislead users into acting on false medical or consequential personal assessments.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 96% confidence

Finding

The file tells the model to make health-related and other sensitive judgments from palm photos but provides no warning, consent language, or limitation notice. In context, that increases the chance users interpret speculative outputs as meaningful guidance, especially for health or intimate life decisions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 96% confidence

Finding

The cited section includes explicit gender-specific prescriptive claims about women, such as framing women as needing '柔顺' and linking certain chart features to moral worth or social inferiority. In a fortune-telling skill, this is not merely historical quotation in isolation: it can be surfaced to end users as authoritative guidance, reinforcing discriminatory stereotypes and potentially causing psychological harm or biased treatment.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 94% confidence

Finding

This content presents physiognomy-based claims linking facial and bodily features to wealth, status, intelligence, misfortune, and lifespan without any framing that it is historical folklore rather than a valid basis for judging real people. In the context of an AI fortune-telling skill, that omission increases the chance the material will be operationalized into discriminatory or demeaning assessments about users or third parties.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The cited passage explicitly prescribes women’s virtue in terms of obedience and chastity as normative doctrine, without framing it as historical material or giving users choice in how such content is presented. In a consumer-facing fortune-telling skill, this can normalize discriminatory gender stereotypes and produce biased or harmful outputs when interpreting female users’ lives, relationships, or worth.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 88% confidence

Finding

All user-facing instructions, trigger phrases, response templates, and required phrasing are written exclusively in Chinese, with no indication that another language can be used. Under the stated policy, a fixed language without user opt-in can be a locale/language policy issue unless explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 87% confidence

Finding

SQP-3 applies to all file types and includes language or locale policy violations. This file presents all instructions and content exclusively in Chinese with no indication that the user can choose another language or that the content is intentionally region- or language-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 83% confidence

Finding

SQP-3 applies to all file types and includes language or locale policy violations. This markdown file presents all content exclusively in Chinese, with no indication that the user can choose another language or that the skill is intentionally limited to a Chinese-speaking context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

This markdown file presents all user-facing content in a single language and does not indicate that Chinese is optional, selectable, or required for a documented region-specific purpose. Under the language/locale policy, forcing a specific language without user opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 88% confidence

Finding

SQP-3 applies to all file types and covers language or locale policy violations. This markdown file presents all instructional content exclusively in Chinese and does not state that the user can choose another language or that the Chinese-only scope is region-specific and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

• Category: Not specified by scanner
• Confidence: 83% confidence

Finding

SQP-3 applies to all file types and covers language or locale policy violations. This file presents all instructional content in Chinese and does not indicate that users may choose another language or that the locale restriction is intentional and justified.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.