Back to skill

Security audit

FitDietKernel

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a genuine diet-tracking skill, but it stores personal health data and uses mutable external code without enough scoping or user control.

Before installing, pin the ClawHub installer version, avoid running install commands as administrator/root, grant the agent access only to the FitDietKernel directory, review profile/log contents for sensitive health data, and require confirmation before any food_registry.json or log writes. Prefer a bundled or pinned Chart.js build with integrity checks, and provide clear controls to delete locally stored weight and diet history.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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 (3)

T08 · Insecure Dependencies

Error
Location
README.md:15
Finding

Execution of an Unpinned npm Installer

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Error
Location
desktop-app.html:7
Finding

Unpinned Third-Party JavaScript Executes in the Application Context

Content
View full analysis
``` The externally loaded script executes in the same page context as the health-data handling code at `desktop-app.html:579-593`, including access to the application's `localStorage`. ### Technical Analysis The application loads Chart.js from jsDelivr without specifying an exact package version. It also omits a Subresource Integrity (`integrity`) hash. Consequently, the content executed by the page can change after review. JavaScript loaded through a normal `
Remediation
View remediation
/dist/chart.umd.min.js" integrity="sha384-" crossorigin="anonymous"> ``` 4. Verify the integrity value against the exact reviewed artifact. 5. Add a restrictive Content Security Policy that limits scripts and outbound connections, preferably using a separate HTTP header when the application is hosted. 6. Avoid inline event handlers and inline scripts so that a strict `script-src` policy can be applied. 7. Clearly document that weight data is stored in browser storage and provide a mechanism to clear it. 8. Correct the storage-key inconsistency so that saved and displayed history uses one intentionally managed key. ]]>

other

Note
Location
README.md:198
Finding

Third-Party Tracking Request Embedded in Project Documentation

Content
View full analysis
``` ### Technical Analysis The README embeds an image hosted by a third-party page-view counter. Documentation renderers that allow remote images will contact `komarev.com` when the README is viewed. This request is unrelated to the core diet-management functionality. It can disclose network and request metadata to the external service, potentially including the viewer's IP address, request time, user-agent information, and referrer information subject to the renderer's privacy controls. No evidence shows that project health data is included in this request. The confirmed issue is the unsolicited third-party contact and resulting metadata exposure. ### Attack Path 1. A user opens the README in a renderer that permits remote image loading. 2. The renderer requests the image from `komarev.com`. 3. The third-party service receives request metadata and associates the request with the project-specific counter identifier. 4. Repeated requests can contribute to usage analytics or viewer tracking. ### Impact Assessment The direct impact is limited to privacy and metadata disclosure. The external service may learn that a client viewed documentation for this project, together with connection metadata permitted by the rendering environment. This finding does not provide local code execution, elevated privileges, or access to the application's stored health records. ]]>
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (30)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared purpose is diet calculation, but the detected behavior includes reading local logs, extracting sleep data, and generating weekly report files. This mismatch is dangerous because users may consent to a nutrition helper without realizing it also processes broader personal activity data and writes derived reports, increasing privacy and data-handling risk.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · desktop-app.html (reported line 401)May include surrounding context.

html
<div class="subtitle">Kernel v1.0</div>
        </header>

        <!-- 今日状态 -->
        <div class="today-card">
            <div class="today-header">
                <div class="day-label" id="dayLabel">周三 · 高碳日</div>

Rp1

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

The README instructs users to run npx clawhub@latest install fitdietkernel, which pulls and executes the latest published package version at install time. Using a moving tag for a package that is executed locally creates a supply-chain risk: if the package or one of its dependencies is compromised, users may run attacker-controlled code during installation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill explicitly tells the AI to generate logs/YYYY-MM-DD.md daily records, creating persistent files on disk, but the README does not clearly warn users that the assistant will write and retain personal health/diet data locally. In this skill context, the stored content may include sensitive lifestyle and body metrics, so silent persistence increases privacy and data exposure risk.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
82% confidence
Finding

The skill instructs the AI to read from a local knowledge-base path and references modifying local data files, but it does not declare any explicit tool scope or permission boundaries. That creates an implicit file-access expectation, which can lead to overbroad local file reads or writes if the hosting agent grants filesystem tools by default.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The usage instructions tell users to modify a local profile file and provide its filesystem path to the AI, but they do not warn about exposing personal health or system-path information. Because profile data and local logs may contain sensitive health metrics and other private details, this can lead to unnecessary disclosure to the model or unintended file access.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
76% confidence
Finding

The only natural-language description in the manifest is written in Chinese, with no indication that users can select another language or that the skill is intentionally region-specific. This may violate language/locale policy if the skill is presented in a multilingual environment without explicit opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill embeds Chinese-only natural-language strings throughout the interface and comments, with no option for user language selection or explanation that the tool is intended only for a Chinese-speaking audience. This can violate language/locale policy when a specific language is forced without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The document is declared as zh-CN and the interface text is written in Chinese, while date formatting is also explicitly forced to zh-CN. This imposes a specific language/locale on users without any visible opt-in or explanation that the skill is intentionally region-specific.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The page loads Chart.js from a third-party CDN at runtime, which introduces a software supply-chain dependency unrelated to a simple personal diet/weight tracker. If the CDN, package, or network path is compromised, arbitrary JavaScript would execute in the app context and gain access to DOM data and locally stored weight history.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The app stores personal health-related data (current weight and history) in localStorage without any visible notice, retention policy, or user consent. localStorage is readable by any script running in the page, so the external CDN script materially increases the exposure of this data if compromised.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill's natural-language UI labels, placeholders, date formatting, and comments are presented in Chinese throughout, with no opt-in, language selection, or documented region-specific justification. This creates a language/locale policy concern because the skill effectively enforces one language for all users.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The file presents all operational instructions in Chinese and uses imperative language such as 'AI 必须严格遵守' without indicating that the user can opt into another language or that the skill is intentionally restricted to a Chinese-speaking context. Under the policy, forcing a specific language without opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly instructs the AI to automatically update food_registry.json from natural-language input, but it does not require explicit user confirmation, validation, or any audit boundary before modifying stored data. In an agent setting, this creates an unauthorized state-change risk: ambiguous or adversarial phrasing could poison the knowledge base, corrupt nutrition calculations, or cause persistent bad recommendations in future interactions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This file contains user-facing natural-language instructions exclusively in Chinese, and there is no opt-in, alternative language, or justification that the content is intended only for a Chinese-speaking audience. Under the policy, forcing a specific language without user choice is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The only user-facing description in the manifest is written in Chinese, which may imply a fixed language/locale experience. The file does not mention that users can choose another language or that the locale restriction is intentional and justified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a skill for fitness and diet knowledge assistance, specifically precise diet calculations such as carb cycling, keto, bulking, and cutting. This file instead reads markdown logs and generates/saves a weekly report branded as "BioContext-Kernel," which is a broader journaling/reporting workflow rather than a direct diet-calculation capability.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The top-level documentation says "BioContext-Kernel Report Generator" and the generated report footer also states it was automatically generated by "BioContext-Kernel". This actively points to a different tool identity and intent than the manifest's fitdietkernel fitness/diet knowledge-base purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring states in Chinese that it automatically converts logs into a structured weekly report, and the generated report strings are also hard-coded in Chinese. This indicates a fixed language/locale behavior with no user opt-in or configurable language selection, which is a natural-language policy violation under the locale rule.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

The entire skill description and example prompt are written only in Chinese, with no indication that users may interact in another language or opt into this locale. Under the policy, forcing a specific language without user choice can be a natural-language policy violation unless the locale restriction is explicitly justified.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The manifest description only says this is a 'personal knowledge base for fitness diet,' which describes a broad topic area rather than specific activation triggers or usage boundaries. In a manifest file, this lack of specificity can make invocation conditions ambiguous and increase the chance of unintended matching for general health or diet-related requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

This code persists user health-related data to fitdiet_data.json, but there is no confirmation prompt, print/log message, or explanatory comment/docstring disclosing that the data is being saved locally. For a code file, local file writes involving personal data should have some visible disclosure unless clearly documented elsewhere.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The code explicitly formats timestamps using the zh-CN locale, which enforces a specific locale presentation regardless of user preference. Under the policy, locale constraints should either be user-selectable or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

When saving weight history, the code serializes dates using toLocaleDateString('zh-CN'), forcing Chinese locale formatting into persisted data. This is another user-visible locale restriction without opt-in or stated justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill saves weight and meal-tracking data to a local JSON file, but the code provides no visible notice, confirmation, or explanatory comment near the write operation. For a code file, file writes affecting personal data should have some form of user disclosure unless clearly documented elsewhere.

Content

No source excerpt is available for this finding.