Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly a cloud-backed child media analysis client, but it handles sensitive child surveillance data and silently creates or reuses identities with too little user control.
Before installing, treat this as a cloud service for sensitive child media, not a local-only analyzer. Use it only with proper consent and a dedicated account, avoid shared workspaces, and verify the provider's retention, access-control, report-sharing, and token-storage practices.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
Backend API communication, token handling, local workspace creation, and reading API key/open-id values from local files are all sensitive behaviors absent from the high-level description. In a skill handling children's surveillance media, undisclosed identity and credential processing meaningfully raises the risk of privacy violations, lateral access, and mishandling of sensitive reports.
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
---
name: "child-emotion-recognition-analysis"
description: "Identifies negative emotions such as crying, anger, fear, and distress through surveillance footage. It actively issues soothing reminders and notifies parents, making it suitable for homes, kindergartens, and daycare centers. | 儿童情绪识别分析技能，通过监控画面识别儿童哭闹、愤怒、恐惧、委屈等负面情绪，主动发出安抚提醒并通知家长，适用于家庭、幼儿园、托管场所等场景"
version: "1.0.19"
license: "MIT-0"
---
# 😊 Child Emotion Recognition Analyzer | 儿童情绪识别分析工具
> **智能健康/识别分析中
The skill description does not clearly warn that child surveillance images/videos and report queries may be transmitted to cloud services. For minors' emotional and behavioral data, lack of upfront privacy disclosure is a serious issue because users cannot give informed consent to processing, storage, and account-linked retrieval of highly sensitive content.
Silently reusing or auto-creating a local default user identity is unjustified for the stated purpose and can cause reports and uploaded child media to be bound to the wrong account without informed consent. On shared systems, this could expose one household or classroom's sensitive history to another user through implicit identity reuse.
The HTTP helper silently auto-registers or logs in users against external service endpoints, caches returned tokens, and reuses them for subsequent requests. In a child-emotion-recognition skill, this is especially risky because operators may believe the skill only processes local surveillance-derived events, while the code actually establishes backend identities and external accounts without clear disclosure or user initiation.
The skill documentation instructs use of shell, network, filesystem, environment, and local persistence capabilities but does not declare any explicit tool scope or permission boundaries. That creates an over-privileged, weakly governed skill where reviewers and runtime policy engines cannot easily constrain what the skill is allowed to access.
The default trigger is broad enough to auto-activate on general mentions of child emotion or uploaded media, which can cause unintended processing of sensitive surveillance footage. Automatic activation is especially risky here because the data concerns minors and may be sent to cloud services or linked to historical accounts without a deliberate user decision.
The skill includes cloud-based historical report retrieval and persistent account association beyond simple emotion recognition. In the context of child surveillance, this expands data retention and discoverability of sensitive historical records, increasing privacy exposure and the consequences of weak authorization.
The history-query auto-trigger uses common phrases that may unintentionally invoke cloud retrieval of sensitive records. In a child-monitoring context, accidental report listing can reveal private historical incidents, report links, and account-associated data to the wrong person or in the wrong conversation.
This is a markdown file, so SQP-2 applies to omissions in safety-relevant documentation. The document states that the API uses API Key authentication and supports file uploads for child emotion recognition analysis, but provides no warning about protecting API keys or the sensitivity of uploaded data.
The script accepts a local file path or remote URL for child emotion analysis and immediately forwards it into the skill pipeline without any user-facing notice, consent step, or data-handling warning. In the context of surveillance footage of children, this creates a real privacy and compliance risk because highly sensitive biometric and behavioral data may be transmitted to backend services or processed in ways the user does not understand.
Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.
result_json = JsonUtil.parse(result_json_pure_text, result_json_pure_text)
result_json_common_ai_response = result_json.get("commonAiResponse") if isinstance(result_json,
dict) else result_json
if result_json_common_ai_response:
result_json = result_json_common_ai_response
Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.
result_json = JsonUtil.parse(result_json_pure_text, result_json_pure_text)
result_json_common_ai_response = result_json.get("commonAiResponse") if isinstance(result_json,
dict) else result_json
if result_json_common_ai_response:
result_json = result_json_common_ai_response
Several user-visible strings are hardcoded in Chinese, such as the analysis report header and export-link message, which forces a specific language experience. The file does not provide any opt-in, locale selection, or justification for restricting output to Chinese.
Detected: suspicious.install_untrusted_source