Security audit
Security checks for vulnerabilities and agentic risk
The skill largely matches a cloud infant-video analysis purpose, but it silently creates or reuses identities, performs remote login/registration, and persists tokens for sensitive child-monitoring data without enough user-facing control or privacy detail.
Review before installing. This skill sends infant sleep media or media URLs to lifeemergence.com cloud services, can query prior cloud reports, silently creates or reuses an internal user identity, and stores authentication tokens in a local SQLite database under the workspace data directory. Install only if that cloud processing and token persistence are acceptable for sensitive child/home footage, and prefer using it with explicit user confirmation before history lookups.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
External HTTP communication, automatic identity/open-id handling, registration/login, token use, and filesystem setup are materially different from the narrow infant-monitoring behavior users are led to expect. This is dangerous because it combines hidden identity processing and remote transmission with a false appearance of a dedicated baby-safety monitor.
This file exposes generic add/edit/delete and arbitrary HTTP verb wrappers that are far broader than the skill’s declared purpose of infant suffocation warning analysis. In a narrowly scoped safety-monitoring skill, such general-purpose remote action capability increases the attack surface and could be reused by other components to send, modify, or delete remote data unrelated to infant safety.
This file defines generic user-account persistence, including user identity records and related lookup/update behavior, which is materially outside the declared infant suffocation warning purpose. In a safety-monitoring skill, hidden or unnecessary account-management code increases the attack surface and creates unjustified collection and retention of personal data unrelated to the advertised function.
The User model stores sensitive profile and authentication-related data including email, birthday, token, and open_token, none of which are clearly required for detecting prone sleeping, head covering, or airway occlusion. Collecting and mutating such data in a child-safety skill creates privacy and credential-handling risk, especially if the database is local, weakly protected, or reused across agents.
This utility file contains open-id resolution, credential discovery, local user lookup, and automatic account creation logic that is unrelated to the declared infant suffocation warning purpose. In a safety-monitoring skill, hidden identity provisioning and persistence materially increases risk because the skill can silently assume or create identities and prepare for backend access beyond what users would expect.
When no identity is present, the code generates a synthetic default user, writes it to the local database, and then reuses it later. Silent creation of persistent identities is unrelated to infant safety monitoring and dangerous because it can establish backend-traceable accounts and durable local state without user knowledge or approval.
The HTTP client performs backend authentication, token loading, token refresh via re-login, and persistence of returned tokens/user records, none of which aligns with an infant sleep hazard detection skill's stated purpose. This broad remote access capability enables silent data transmission and account lifecycle actions, making the skill context more dangerous because users would reasonably expect local safety analysis, not background identity-backed API operations.
This request path sends usernames, open IDs, tenant data, skill metadata, and authentication tokens to remote services without any visible user-facing disclosure or opt-in. In a baby-safety alerting skill, hidden authenticated network transmission is especially concerning because the functional expectations are narrow and safety-focused, making undisclosed backend communication disproportionately risky.
The skill documents and encourages shell execution, file handling, network access, and local state manipulation, but it does not declare any explicit tool-scope restrictions such as allowed tools or permissions. This creates an overbroad execution surface where the agent may use more capabilities than the metadata communicates, increasing the risk of unintended command execution, data access, or exfiltration.
The historical-report auto-trigger keywords are broad and may cause unintended cloud queries when users mention general history-related phrases. This can lead to unnecessary disclosure of sensitive prior infant-monitoring reports or accidental retrieval of another user's associated data if identity binding is weak.
The skill asks users to provide local files or remote URLs for analysis and states that API services will download network resources, but it does not prominently warn users that media may be transmitted to cloud services. In the context of infant sleep videos, this is privacy-sensitive because uploads may contain highly sensitive footage from inside a home and of a child.
The file is largely bilingual, but the operational instructions and output formatting requirements are primarily prescribed in Chinese, including trigger phrases and reporting guidance, without explicitly offering the user a language preference. This may amount to a language/locale constraint that is not presented as optional or justified as region-specific.
The documented APIs are for pet health analysis, which conflicts with the stated purpose of an infant suffocation warning skill. This kind of cross-domain mismatch can cause the skill to call or expose unrelated backend capabilities, leading to incorrect processing, data mishandling, or deployment of the wrong service in a safety-critical context.
Detected: suspicious.install_untrusted_source
