Back to skill

Security audit

菲菲老师学习主控

Security checks for vulnerabilities and agentic risk

Overview

This education skill is not malicious, but it asks agents to track and sync sensitive student information without enough consent, storage, or safety boundaries.

Install only if you are comfortable with a family education assistant tracking a child's learning profile and generating parent-facing reports. Before using it with real minors, require explicit guardian/student consent, disable or clarify cloud sync, avoid storing school/location/schedule unless necessary, and treat emotional-support output as educational support rather than professional mental-health care.

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
references/knowledge_base.md:493
Finding

Unspecified Local Persistence and Cloud Synchronization of Sensitive Student Profiles

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (11)

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 6)May include surrounding context.

md
---
name: feifei-learning-commander
version: "3.5.1"
author: "学来学去学习社 | Xue Lai Xue Qu Learning Society"
brand: "学来学去出品,必是教育精品 | Excellence in Education"
description: "【学来学去学习社出品】菲菲老师学习主控 v3.5 - 中国家庭AI超级助手 · 全科总指挥+知识网络总控+亲子翻译官+成长陪伴师 | Feifei Teacher Learning Commander v3.5 - China's AI Super Assistant for Families. Learning Commander + Knowledge Network Control + Parent-Child Translator + Growth Companion. 给不一样的孩子,拥有不一样的人生!"
metadata:
  clawdbot:
    emoji: "🌸"
    tags: ["菲菲老师", "Feifei Teacher", "学习主控", "Learning Commander", "知识网络", "Knowledge Network", "亲子沟通", "Parent-Child Comm

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The README promotes proactive monitoring of learning progress and parent communication, which implies collection and ongoing use of potentially sensitive student data, possibly including minors' educational profiles. In an education/family context, omitting clear disclosure, consent, and data-handling boundaries can mislead users into enabling surveillance-style features without understanding privacy consequences.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · README.md (reported line 94)May include surrounding context.

text

菲菲老师会自动执行以下流程:
Feifei automatically executes:

🧠 识别: 初二数学,人教版,第14章 一次函数

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The advertised 'heartbeat service' describes active reminders, task assignment, completion checking, and repeated follow-up, which amounts to persistent behavioral monitoring and unsolicited intervention. For a family education skill aimed at children, this is risky without clear notice, consent, frequency controls, and an easy way to disable the feature.

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 psychological support, emotional guidance, and family-education assistance for children without clear user-facing limits, crisis disclaimers, or escalation guidance. In a family/child context, users may over-rely on the agent for sensitive mental-health or parenting decisions, creating risk of harmful advice, delayed professional care, or inappropriate handling of distress disclosures.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill describes progress monitoring, knowledge-network updates, and parent-facing reports about a child, but provides no notice about what data is collected, retained, shared, or how consent is handled. Because the data concerns minors and family relationships, omission of privacy and data-handling boundaries increases the risk of oversharing sensitive educational and behavioral information to parents or third parties.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

This manifest file uses a broad descriptive phrase for the skill but provides no explicit trigger phrases, activation conditions, or exclusions. In a manifest, the lack of specificity can make invocation behavior ambiguous and increase the chance of unintended activation in general educational or family-assistant contexts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The document is written primarily for a Chinese educational setting, with Chinese-first scripts, Beijing region defaults, Chinese textbook mappings, and exam-planning assumptions tied to the Chinese school system. Because the file does not state that this locale/language scope is optional or user-selected, it risks violating the policy against forcing a specific language or locale without opt-in.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document explicitly defines automatic local writes, daily cloud synchronization, and retention of recent historical versions for student profile data, which includes identifiable and sensitive educational information. In a family/child-focused assistant context, undisclosed persistence and cloud transfer materially increase privacy, consent, and compliance risk, especially for minors.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
71% confidence
Finding

The natural-language description presents the skill as "China's AI Super Assistant for Families," and the name is primarily Chinese, which suggests a locale-specific orientation. Because the manifest does not state that users can choose language/locale or explain a justified region-specific constraint, this may conflict with language/locale neutrality expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The text says all templates support only Chinese/English bilingual output, which sets a specific language policy for the skill output. Under the policy rule, forcing a specific language or locale without user opt-in is a natural-language policy concern unless the user is offered a choice or the restriction is clearly justified.

Content

No source excerpt is available for this finding.