Security audit
Security checks for vulnerabilities and agentic risk
The skill is not overtly malicious, but it needs Review because it handles sensitive rehabilitation audio/video through cloud services while silently managing identities, tokens, history, and escalation actions.
Install only in an environment where patients or authorized caregivers have explicitly consented to cloud processing of rehab video/audio, identity linkage, history retention, and any therapist or family notifications. Before use, verify the lifeemergence.com endpoints, retention policy, token storage location, alert recipients, and whether automatic escalation can be disabled or put behind human approval.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
The skill appears to perform outbound API communication, create or look up user identities, read/write workspace files, and manage agent context, none of which are clearly disclosed by the narrow rehab-analysis description. Because the content concerns sensitive patient video and history, undisclosed identity and data-handling behavior materially increases privacy and trust risk.
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
---
name: "smyx-rehab-motivation-encouragement-analysis"
description: "Through fixed cameras in rehabilitation centers or home rehab areas, the system analyzes video of patients during rehabilitation training to detect frustration / giving-up tendency behaviors: sighing (rapid chest-abdomen rise-fall with exhalation), training interruption (actively stopping before reaching preset reps or duration), head-down silence (head lowered, avoiding eye contact, long-term silence), sluggish or. | 通过康复中心或家庭康复区的固定摄像头，分析患者在进行康复训练时的视频，检测沮丧/放弃倾向行为：叹气（胸腹快速起�
The skill handles highly sensitive video/audio monitoring, cloud report queries, and automatic identity association, but the description does not provide a prominent upfront warning about those privacy-relevant behaviors. In a rehab setting, absence of clear notice undermines informed consent and can lead to unexpected surveillance or data disclosure involving vulnerable patients.
This file exposes broad generic network and CRUD helpers (add, edit, delete, http_get/post/put/delete) that can send arbitrary requests to caller-supplied URLs, which substantially exceeds the declared purpose of analyzing rehab-training video for patient motivation signals. In an agent/skill context, such overbroad capabilities increase the attack surface for unauthorized data access, remote manipulation, and misuse as a general-purpose network client, especially because the skill operates in a sensitive healthcare context.
The methods at these lines provide arbitrary remote resource interaction and mutation through generic POST/PUT/GET/DELETE primitives with caller-controlled URLs. In a skill ostensibly limited to behavioral analysis of rehab patients, this unjustified remote manipulation capability is dangerous because it could be repurposed to alter external systems, exfiltrate sensitive medical-related data, or pivot to other internal or third-party services.
This code provisions user identities, reads an identity file from the workspace, persists local account state, and selects a current open-id automatically. Those capabilities are unrelated to the stated video-behavior analysis purpose and create unauthorized identity handling and account lifecycle behavior that could expose or misuse patient/operator identities.
The code automatically reads a workspace identity file and, failing that, creates synthetic default user identities and stores them locally. For a medical-adjacent rehab analysis skill, silently assigning identities is dangerous because it can misattribute actions/data, undermine auditability, and process sensitive workflows under fabricated or unintended accounts.
The HTTP utility automatically contacts external APIs, injects account and platform metadata, retries authorization, and persists returned tokens. In the context of a rehab motivation analysis skill, this is materially beyond scope and enables silent external data transmission and remote account interactions without a clear need tied to local video analysis.
This block auto-registers or logs in users against a remote health endpoint using derived usernames/openIds, without any indication that the skill's function requires remote account creation. In the rehab context, that creates unjustified remote exposure of identity-related data and could establish accounts or sessions on behalf of users without informed consent.
The skill advertises and instructs use of shell, file, network, and environment-backed capabilities but does not declare any explicit tool scope or allowed-tools boundary. In a skill that handles sensitive rehab video and cloud queries, this increases the chance of unintended or overly broad capability use beyond what users would reasonably expect.
The documented behavior goes beyond detection into active intervention, escalation, and therapist report delivery, but this broader operational scope is not cleanly reflected in the core declared purpose. In a sensitive healthcare-adjacent workflow, under-disclosing intervention capabilities can mislead reviewers and users about the real-world effects and data sharing caused by the skill.
The skill expands from behavior analysis into escalation that notifies family members, which is a materially different and more sensitive action than simple monitoring. In a rehab context, family notification can expose health-related behavioral information to third parties without sufficiently clear consent, policy controls, or access checks.
The default trigger activates on broadly defined rehab video analysis requests, which can cause the skill to run in situations where users did not intend sensitive frustration monitoring, identity-linked history behavior, or cloud processing. For health-related media, broad auto-invocation increases the chance of collecting or transmitting sensitive data without sufficiently specific user intent.
The documentation states that the system will automatically create or reuse an internal default user identity, even when the user did not explicitly provide one. In a skill handling sensitive rehabilitation media and historical reports, silent identity creation/linkage can cause cross-session data mixing, unauthorized association of records, and reduced user awareness of how patient data is being tracked.
The manifest says this skill analyzes rehabilitation patients in rehab-center or home-rehab videos for frustration or giving-up behaviors. In contrast, the implementation retains animal-oriented semantics via pet_type, offers cat and dog as valid categories, and mutates a default pet-type setting, indicating code behavior or lineage that does not match the stated human-rehabilitation purpose.
Detected: suspicious.install_untrusted_source