Back to skill

Security audit

weight-loss-coach

Security checks for vulnerabilities and agentic risk

Overview

This weight-loss coaching skill is broadly coherent, but it tells the agent to permanently remember and accumulate sensitive health and lifestyle data without clear consent, retention, or deletion controls.

Install only if you are comfortable with the agent remembering health and diet information across sessions. Before using it, ask for session-only handling or confirm how to review, delete, or disable stored health data; avoid uploading screenshots with unrelated personal details unless you intend them to be processed.

Vulnerability Patterns
  • 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
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:142
Finding

Indefinite Retention and Aggregation of Sensitive Health Data Without Privacy Controls

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (12)

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill’s activation scope is extremely broad and includes common, ambiguous everyday phrases, which can cause unintended invocation in unrelated conversations. In this context, accidental activation is more dangerous because the skill is designed to collect and retain sensitive health, diet, and preference data, so a misfire can lead directly to unnecessary processing of sensitive information.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill explicitly instructs proactive tracking, memory, and retention of health-related data and preferences across sessions without a clear notice or consent flow. This is dangerous because the data involved includes weight, age, sex, activity level, food intake, and health screenshots, all of which are sensitive and can create privacy harm if stored or reused unexpectedly.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The description and all prescribed interactions are written as if the skill should operate in Chinese, including Chinese trigger phrases and Chinese-platform-specific behavior. There is no opt-in or user choice for language/locale, and no explicit statement that the skill is intentionally limited to a Chinese-speaking region or audience.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The instruction to proactively track user data and remember preferences across sessions creates a natural-language pathway for persistent retention of sensitive health information. In this skill, that retention is especially risky because broad activation increases the chance that sensitive data is collected and later surfaced or reused outside the user’s expectations.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The profile-building flow collects and reuses highly sensitive attributes such as height, weight, age, sex, activity level, and goal weight, then instructs the system to remember them long-term. This creates a privacy and data-minimization problem because the skill normalizes permanent retention of sensitive health-adjacent profile data without explicit necessity or user control.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill treats health screenshots as inputs to extract and summarize sensitive data such as heart rate, sleep, body composition, and exercise metrics. This is dangerous because screenshots often contain more information than needed, and without explicit minimization or consent the skill may process or retain a broad set of personal health signals.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The conversation policy explicitly mandates permanent memory and cumulative logging of health/activity history, which is a direct retention risk for sensitive personal data. In a weight-loss context, longitudinal records of weight, intake, exercise, and goals can reveal intimate health patterns and create downstream exposure if reused, leaked, or surfaced unexpectedly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The recipe and takeout trigger list is also overly broad, with phrases like '今天吃什么' or '推荐一下菜谱' that may arise in general conversation without consent to enter a tracking-oriented health workflow. Because this skill couples recipe suggestions with calorie accounting and personalization, ambiguous activation can pull users into sensitive profiling they did not clearly request.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill directs use of web_search for recipe discovery without warning that user query context may be transmitted to external services. In a health-management skill, even recipe-related queries can reveal dieting status, preferences, or other sensitive context, making silent external sharing a meaningful privacy risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file contains user-facing instructions exclusively in Chinese, and there is no indication that the skill is region-specific or that users can opt into another language. That can violate a language/locale policy requiring user choice or documented justification for a fixed locale.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The entire markdown file is written only in Chinese, from the title through all instructions and labels, with no indication that language selection is optional. Under the policy rule, a skill artifact that imposes a specific language without opt-in can be a natural-language locale violation unless the regional limitation is explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

SQP-3 applies to all file types and covers language or locale policy violations. This markdown file presents all instructions and reference content only in Chinese, with no indication that users can choose another language or that the locale restriction is intentional and documented.

Content

No source excerpt is available for this finding.