Security audit

Fitness encyclopedia

Security checks across malware telemetry and agentic risk

Overview

This is a real fitness helper, but it needs review because it can activate too broadly and asks for sensitive body, diet, and injury details without enough privacy or medical-safety framing.

Install only if you are comfortable with a fitness skill activating on broad exercise-related language and asking for personal body, diet, and injury information. Treat nutrition, injury, joint, and rehabilitation suggestions as general fitness information, not medical advice, and avoid sharing sensitive health details you do not want handled by the assistant.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (13)

Intent-Code Divergence

Low

Confidence: 96% confidence
Finding: The validation path attempts to terminate with sys.exit(1), but sys is never imported. If invalid input is supplied, the script will raise a NameError instead of cleanly handling the error, causing a denial-of-service style crash and breaking the documented safety check path. In this fitness-calculation context, the impact is limited to reliability and predictable input handling rather than code execution or data compromise.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The documented automatic triggers are very broad terms like 'fitness', 'workout', 'training', and 'exercise', which are common in normal conversation. This can cause unintended activation of the skill, leading users to disclose health or body-related information without clearly intending to engage this tool.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill requests sensitive personal and health-related information including gender, height, weight, age, body fat percentage, joint limitations, and dietary restrictions, but the README provides no privacy notice, retention policy, or data handling explanation. In a health-oriented skill, this increases the risk of users sharing sensitive data without informed consent or understanding how it will be stored or processed.

Vague Triggers

High

Confidence: 97% confidence
Finding: The catch-all trigger activates on essentially any fitness-related language, making routing overly aggressive and likely to hijack benign or ambiguous conversations. This can suppress normal intent clarification, cause the wrong skill to take control, and steer users into a rigid workflow they did not request.

Vague Triggers

High

Confidence: 98% confidence
Finding: Mandatory triggering on generic keywords like 'fitness', 'training', or 'exercise' is unsafe because these terms appear in many harmless contexts and do not reliably signal the user's intent to invoke this skill. Overbroad activation can cause inappropriate skill takeover, reduce routing integrity, and increase the chance that sensitive health guidance is delivered without proper context.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill solicits personal and health-related data including gender, height, weight, age, body fat percentage, and joint limitations without any privacy notice, minimization, or explanation of handling. This creates unnecessary exposure of sensitive information and can pressure users into disclosing data before they understand why it is needed.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The manifest trigger list is very broad and consists of common fitness-related terms such as 'fitness', 'workout', 'training', and 'exercise' that are likely to appear in many ordinary user requests. This can cause the skill to activate unintentionally, increasing the chance that users are routed into this package without clear intent and expanding the attack surface for any downstream unsafe behavior in the skill.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The manifest trigger list is very broad and consists of common fitness and health terms such as 'fitness', 'workout', 'training', and 'gym', which are likely to appear in many ordinary user conversations. This can cause the skill to activate in situations far beyond explicit user intent, increasing the chance of unwanted invocation and over-collection or mishandling of sensitive health-related context.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The manifest trigger list is very broad and includes common everyday fitness terms such as 'fitness', 'gym', 'exercise', and 'training', which can cause the skill to activate in many benign conversations that only loosely relate to its purpose. In a health-related skill, unintended invocation is risky because the agent may provide personalized nutrition or training guidance in contexts where the user did not explicitly request this skill, increasing the chance of inappropriate advice, privacy overreach, or confusing task hijacking.

Missing User Warnings

Low

Confidence: 86% confidence
Finding: The document gives specific calorie-burn figures, heart-rate guidance, and training recommendations across the full guide, but only places caveats and data limitations at the very end. In a fitness/advice context, users may act on the numbers and prescriptions as authoritative guidance without seeing early warnings that the values are approximate and not a substitute for personalized or medical advice.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The document gives specific intake ranges, macro ratios, portion equivalents, and meal-timing advice that users could reasonably treat as prescriptive health guidance, yet it lacks a prominent safety disclaimer for people with medical conditions, allergies, eating disorders, pregnancy, kidney disease, diabetes, or other special nutritional needs. Although the content is not overtly dangerous, omission of context and individualized caution can lead to inappropriate diet choices and health harm when used as a decision aid.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The guide instructs the agent to collect body and health-related personal data such as sex, height, weight, age, and body-fat percentage, but provides no notice about why the data is needed, how it will be used, whether it will be stored, or whether providing it is optional. In a fitness-coaching context this creates avoidable privacy risk and can lead users to disclose sensitive data without informed consent or minimization.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The skill asks users to disclose joint limitations and injury-related information, which is health data, without warning that medical/health information is being collected or clarifying the limits of the advice. This is risky because users may reveal sensitive condition data without informed consent, and the system may be seen as soliciting quasi-medical information without appropriate safeguards.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.