Security audit
Security checks for vulnerabilities and agentic risk
The skill's main purpose is coherent, but it silently creates or reuses an internal identity, stores tokens locally, and sends pet media and report queries to external cloud APIs with limited user control.
Review this skill before installing. Use it only if you are comfortable sending pet images/videos, URLs, and vaccination-report lookups to the configured lifeemergence cloud services, and if silent internal identity creation plus local token storage in the workspace is acceptable for your environment. Prefer an isolated workspace and confirm consent, endpoint ownership, retention, and account cleanup expectations.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The reported behavior includes authenticated outbound requests, automatic user resolution/default user creation, local database operations, and filesystem access, which are substantially broader than the advertised vaccination-reminder use case. In a skill that handles media uploads and historical reports, undisclosed identity and storage operations increase privacy risk and make abuse harder to detect.
The code implements a generic video analysis/history CLI, not the declared vaccination-reminder workflow involving pet identity verification and vaccination-record comparison. This capability mismatch is dangerous because users and integrators may trust the manifest and supply sensitive pet/owner media or rely on outputs for operational decisions, while the actual behavior sends data to an unrelated analysis path and produces misleading results.
The implemented add, edit, and delete methods enable record creation, modification, and deletion even though the skill description says it only compares vaccination dates and returns reminders. That mismatch is dangerous because it violates least privilege and could allow accidental or intentional tampering with vaccination or other hospital records if these methods are reachable through the skill's execution path or shared codebase.
The file exposes generic add/edit/delete methods and arbitrary http_get/http_post/http_put/http_delete wrappers that are not constrained to the vaccination-reminder workflow. This creates a broad capability surface that exceeds the declared read-only, database-comparison purpose and could be reused by the skill or adjacent code to access or modify unrelated internal resources, especially in a hospital-management context handling sensitive records.
The HTTP helper can silently auto-register or log in users against an external health endpoint via /sys/phoneLogin, then cache returned tokens for later use. For a pet vaccination reminder skill, this is over-privileged and dangerous because it can create or assume identities without disclosure, enabling unauthorized external account activity and access expansion.
The skill advertises and instructs use of shell, filesystem, network, and environment-backed operations, but does not declare an explicit tool/permission scope. That increases the blast radius because an agent may execute powerful capabilities without clear least-privilege boundaries or user-visible authorization constraints.
The skill states that uploaded attachments are automatically saved locally and that URLs/media are processed through cloud/API services, but it does not provide a clear user warning in the high-level description about storage and external transmission. This weakens informed consent and can expose sensitive pet-owner-associated media and metadata to retention or third-party processing risks.
The documentation exposes endpoints for face-recognition health analysis and full report export in a workflow involving biometric identification and vaccination records, which are sensitive personal data categories. Even though this is only documentation, omitting any warning, access-control expectations, minimization guidance, or handling constraints increases the risk that integrators will build unsafe flows, overexpose report export capability, or process sensitive data without adequate consent and safeguards.
The comment at L02 is written only in Chinese and there is no indication in this file that language selection is optional or configurable. Under the policy, forced single-language behavior or language-specific instructions without opt-in can be a locale/language policy concern when not clearly justified.
Line L17 returns a user-facing string entirely in Chinese, with no indication that the user can choose another language or locale. For a code file, this is a natural-language policy concern because it appears to enforce a specific language in output.
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
Detected: suspicious.install_untrusted_source