Security audit
Security checks for vulnerabilities and agentic risk
The skill appears to perform the advertised cloud posture-video analysis, but it also silently creates or reuses identities, stores tokens locally, and can query cloud history for workplace camera data without enough user-visible scoping.
Review this before installing in any workplace environment. Use it only if employees have given informed consent, the lifeemergence.com cloud service is approved for camera footage and posture records, and you are comfortable with local account/token persistence in the workspace data directory. Consider requiring explicit confirmation before history queries or video uploads.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
Network communication, local file access, identity resolution, and database interaction are all reported, yet none of the promised video-analysis or reminder logic exists in code. This combination is particularly dangerous because it resembles a data-collection and account-linkage workflow disguised as a simple office wellness tool.
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
---
name: "smyx-office-worker-posture-warning-analysis"
description: "Using a fixed camera in the office (aimed at the workstation), the system analyzes office workers' sitting-posture video in real time, detecting continuous sitting duration, neck forward angle (head offset relative to shoulders), and back curvature (hunchback degree). | 通过办公区固定摄像头（对准工位）实时分析办公人员的坐姿视频，检测连续坐姿时间、颈部前倾角度（头部相对于肩部的偏移）、背部弯曲度（驼背程度）。当久坐时间超过预设阈值（默认1小时）且未起身活动时，输出'久坐提醒'；当颈
The User model stores sensitive fields including token, open_token, email, birthday, age, and realname without any justification tied to posture analysis. In the context of a camera-based workplace monitoring skill, this is especially risky because it combines behavioral surveillance with identity and authentication-like data, increasing harm from misuse or database compromise.
The code reads workspace identity material from data/smyx-api-key.txt and falls back to creating persistent local user records and default identities. For a posture-warning camera skill, this is unrelated functionality that can silently appropriate local identity state and establish durable tracking across sessions.
The utility layer performs remote login/account creation and token acquisition even though the declared skill purpose is local office-posture video analysis. This creates undisclosed identity provisioning and outbound data flows, expanding the trust boundary and enabling silent enrollment of users or systems into external services.
The skill declares broad operational behavior that implies shell, filesystem, network, and environment access, but it does not explicitly constrain tool permissions in the manifest. In an agent environment, missing scope declarations can allow the skill to exercise more capability than users or reviewers expect, increasing the blast radius of prompt-triggered actions.
While the file contains some bilingual overview text, the actionable trigger rules, workflow, constraints, notes, and examples are predominantly in Chinese. This imposes a language preference on users and operators without stating that the skill is Chinese-locale specific or offering an explicit language option.
A default trigger for generic workstation video analysis can cause the skill to run on broader inputs than users intended, including potentially sensitive office footage. Overly broad auto-invocation raises the chance of accidental data submission to external services and bypasses meaningful user confirmation.
The manifest presents a limited real-time posture-warning skill, but the body expands it into cloud-backed historical report querying and report-link retrieval. This scope expansion matters because it introduces additional data access paths and retention concerns that users may not expect from the top-level description.
Broad natural-language triggers for historical report queries can invoke cloud data retrieval without sufficiently clear user scope or authorization checks. In a workplace-monitoring context, accidental or overbroad history retrieval can expose sensitive employee behavioral records.
The user-facing string at L17 is hard-coded in Chinese, which imposes a specific language/locale on all users. The file does not indicate any opt-in, fallback, or region-specific justification for this restriction.
The skill is presented as office-worker posture analysis, but it exposes an unrelated 'pet_type' parameter and mutates a global default based on user input. This mismatch is dangerous because it indicates code reuse or hidden multi-purpose behavior, which undermines trust in what data is being processed and may cause the backend to route requests in unintended ways.
The code includes a hidden open-id based listing capability and internal identity handling that are unrelated to the stated posture-analysis purpose. Hidden account-scoped retrieval features increase the risk of unauthorized access to prior analysis records or user data, especially because the capability is not transparently disclosed in the main skill description.
The tool accepts a network video URL for analysis but provides no meaningful warning that office camera footage may be fetched from or transmitted to external services. In a workplace-monitoring context, this is sensitive personal data, so lack of disclosure can lead to privacy violations, policy breaches, or accidental submission of internal surveillance footage to third parties.
Detected: suspicious.install_untrusted_source