Back to skill

Security audit

育儿助手

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a parenting helper, but it can automatically write vaccine appointments and persistent reminders without clear confirmation.

Review before installing. The parenting reference content is disclosed, but do not let it automatically create vaccine calendar entries or cron reminders unless you are comfortable with persistent local reminders containing child health details. A safer version should ask for confirmation, show the exact event/reminder details, validate vaccine fields, and avoid shell command construction.

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

Error
Location
SKILL.md:114
Finding

Shell Command Injection Through Unsanitized Vaccination Data

Content
View full analysis
(第<剂次>针)" \ --date "" \ --time "10:00" \ --duration 60 \ --notes "<疫苗说明>" \ --calendar "育儿提醒" ``` ### Technical Analysis The Skill instructs the Agent to insert a vaccine name, dose number, date, and notes into a shell command. These values can originate from user input, but the documented flow does not require validation, sanitization, or shell-safe argument handling before execution. Double quotation marks do not make arbitrary input safe when an Agent constructs and executes a shell command. An attacker could provide input containing a closing quotation mark, command substitution such as `$(...)`, backticks, or other shell metacharacters. If the Agent incorporates that value into the command as documented, the shell can interpret attacker-controlled syntax rather than treating the entire value as inert data. The external calendar script is not included in this project and was therefore outside the reviewed code. The vulnerability arises before that script processes its arguments: the shell may evaluate injected syntax while parsing the command. ### Attack Path 1. An attacker submits a vaccination statement containing a crafted vaccine name or description. 2. The Skill directs the Agent to calculate the next dose and automatically invoke the calendar command. 3. The Agent substitutes the attacker-controlled text into the `--title` or `--notes` value. 4. The Agent sends the constructed command to a shell without robust escaping or an argument-array process API. 5. The shell evaluates the injected command substitution or metacharacters. 6. The injected operating-system command runs with the same permissions and environ ...[truncated 1048 chars]
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 (10)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest promises automatic calendar and reminder creation whenever the user mentions a vaccine event, but it does not require explicit consent before performing those side effects. This is dangerous because a casual statement about a past vaccination could cause writes to the user's calendar and local reminder/scheduling systems without clear approval.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The operational instructions explicitly require the agent to add a calendar event and create a cron-based reminder as mandatory steps, with no user-confirmation checkpoint before those side effects. In context, this is more dangerous because the skill handles health-related information about a child and writes to external/local systems, creating risks of unauthorized actions, privacy leakage, and persistent unwanted reminders.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill presents health and fever-handling guidance for infants and young children without an explicit warning that the information is general educational content and not a substitute for professional medical care. In a pediatric context, users may over-rely on the assistant during urgent situations, delay seeking care, or apply advice inappropriately to a child with red-flag symptoms.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest description uses very broad parenting-related trigger terms, so ordinary conversation about babies, sleep, food, or education could invoke the skill unexpectedly. Because this skill includes operational behaviors such as scheduling calendar events and reminders, accidental activation increases the chance of unintended side effects and privacy-sensitive processing.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This file gives pediatric home-care and medication guidance, including fever thresholds, ORS use, temporary formula changes, montmorillonite powder, topical steroids, and seizure response, but lacks a clear disclaimer that it is not a substitute for professional medical evaluation. In a baby-care skill, users may rely on this advice to delay care or self-medicate incorrectly, which raises the risk of harm for infants and young children, a population with lower safety margins and faster deterioration.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The entire file is written in Chinese and provides no indication that the user can choose another language or locale. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README says users should 'directly ask in Chinese,' which imposes a language constraint in natural-language instructions. There is no opt-in language choice or explanation that the skill 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
95% confidence
Finding

All user-facing content in the file is presented only in Chinese, which imposes a language choice without opt-in. Under the policy, language constraints should either offer user choice or be clearly documented as region- or audience-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file contains all user-facing guidance in Chinese, and there is no indication that the skill is intentionally region-specific or that users can opt into another language. Under the stated policy, forcing a specific language without user choice 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
84% confidence
Finding

The entire skill content is presented only in Chinese, with no indication that language selection is optional or that the content is intended solely for a Chinese-speaking or region-specific audience. Under the policy criteria, a skill that fixes a language or locale without opt-in can be considered a natural-language policy violation.

Content

No source excerpt is available for this finding.