Security audit
Security checks for vulnerabilities and agentic risk
This skill mostly matches its pet-ear analysis purpose, but it silently creates or reuses identities, stores tokens locally, and can fetch cloud report history with broad automatic triggers.
Install only if you are comfortable sending pet ear images/videos or URLs to the configured Life Emergence cloud service and having the skill create or reuse an internal account identity. Review how to delete local workspace data and service-side report history before using it with sensitive media or shared machines.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
The skill appears to perform remote HTTP communication, token/auth management, persistent user/open-id generation, and workspace filesystem discovery without clearly declaring those behaviors in the core purpose. Those hidden platform and identity operations increase the blast radius well beyond simple veterinary media analysis.
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
---
name: "smyx-ear-health-snapshot-analysis"
description: "Triggers when a user provides a pet ear/scratching/head-shaking video URL or file for analysis; uses smart camera to monitor head-shaking and ear-scratching actions, automatically captures HD ear-canal images, and applies AI vision analysis to identify ear-canal color (pink/red/dark red), presence of black granular discharge (ear mites), and degree of earwax accumulation. When redness, large discharge or suspected ear mites are detected, outputs abnormality alerts and recommends owner check-up or veterinary visit. Application: pet families, boarding centers, pet hospital pre-scre
Automatic internal identity initialization, fallback account reuse, and account creation are unrelated to the user-facing promise of pet ear analysis and create silent identity binding. This can cause one user's actions to be linked to the wrong persisted identity or enable access to prior records without transparent authentication semantics.
The implemented code does not perform the ear-health video or image analysis described in the skill metadata and instead contains a stubbed ai_chat method with commented-out subprocess execution and broken result handling. This mismatch is dangerous because downstream systems or users may rely on the skill for health-related screening, but it silently provides no real analysis, creating false assurance and potentially delaying veterinary care.
This utility code creates, resolves, and persists user identities unrelated to the stated ear-health image analysis function, including reading a local key file and creating fallback local identities. In the context of a pet ear-analysis skill, hidden identity provisioning and persistence materially expands data collection and account linkage without clear necessity or user awareness.
The HTTP helper automatically logs in or registers users against external services, injects tokens into requests, and stores returned tokens locally. For an ear-health visual analysis skill, this is over-privileged behavior that can silently bind users to external accounts and enable broad data transmission beyond the narrow expected function.
The skill declares powerful capabilities in its documented workflow—shell execution, network access, file read/write, and likely environment access—but does not scope or declare them in manifest metadata. That weakens least-privilege enforcement and makes it harder for a host to constrain what the skill may do, especially since it instructs execution of local scripts and cloud queries.
The default trigger conditions are broad enough to auto-invoke the skill on ordinary mentions of ear symptoms plus file uploads. Overbroad triggering can cause unintended processing of files, unnecessary remote API calls, or accidental collection/storage of media without sufficiently explicit user intent.
Automatic cloud-based historical report querying expands the skill from one-time media analysis into remote record retrieval. That increases privacy exposure and the chance of overbroad data access, especially when the feature is triggered by common phrases and tied to internal identity handling.
The history-query trigger phrases are common and can overlap with normal conversational requests, increasing the risk of unintentional cloud record retrieval. Because history access can expose prior reports, accidental activation is more sensitive than accidental analysis.
The documentation says uploaded attachments are automatically saved as local files, which broadens data retention and local exposure compared with transient analysis of supplied media. Saving user-provided files locally can leak sensitive media through insecure permissions, leftover artifacts, or unintended reuse.
The instruction mandates a specific Markdown table output format with Chinese column labels such as '报告名称', '宠物类型', '分析时间', and '点击查看'. This imposes a locale/language choice in the skill behavior without stating user opt-in or offering alternatives.
The method returns a fixed Chinese-language output heading string, which imposes a specific language on all users. The file does not indicate any user opt-in, locale selection, or region-specific justification for this language constraint.
The script exposes a history-report listing mode via --list and show_analyze_list() even though the manifest describes only analysis of a supplied video or URL. Undocumented access to prior reports expands the data-access surface and can reveal historical pet health results associated with a user identity, which is sensitive operational/user data even if not traditional medical data.
The code resolves an internal OpenID and then uses ConstantEnum.CURRENT__OPEN_ID to fetch analysis history, despite the skill’s stated purpose being single-video ear analysis. Tying hidden identity resolution to report access creates a risk of unauthorized data retrieval or cross-user history exposure if identity resolution is weak, implicit, or attacker-influenced.
OpenIdUtil.resolve_current_open_id(args.open_id, use_current=bool(args.open_id)) initializes internal identity handling without any user-facing disclosure, and the associated CLI parameter is hidden from help. Hidden identity processing is dangerous because users may unknowingly trigger access to account-scoped data or cause actions to execute under an internal identity context they did not explicitly consent to.
Detected: suspicious.install_untrusted_source