Security audit
Security checks for vulnerabilities and agentic risk
The skill is not clearly malicious, but it sends pet-area media and identity-linked report activity to a cloud service while silently creating or reusing an account identity.
Install only if you are comfortable with litter-box-area videos or URLs, derived reports, and identity-linked history being sent to the configured Lifeemergence cloud service. Use non-sensitive clips, avoid footage that reveals people or private interiors, and be aware that the skill can create/reuse a remote account and store local tokens for later report access.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The skill claims specialized litter-box video analytics and health-alert behavior, but the supplied behavior appears to be generic API wrappers, user/account handling, and report retrieval. This mismatch is dangerous because reviewers and users may authorize sensitive video, cloud access, and identity-linked history operations under false assumptions about what the skill actually does.
The manifest was flagged for tool/metadata poisoning indicators, and this skill already contains strong behavioral instructions about mandatory script execution, cloud querying, and hidden identity handling. In agent ecosystems, misleading or manipulative manifest metadata can cause the orchestrator or reviewer to trust unsafe capabilities, especially when combined with undeclared network, shell, and user-association behavior.
---
name: "smyx-litter-box-usage-monitor-analysis"
description: "Triggers when a user provides a litter-box area video URL or file for analysis; uses object detection and tracking to identify each cat's entry/exit times at the litter box, records daily usage frequency and per-visit duration (entry → exit), and compares against the historical baseline. When frequency rises/falls significantly, or per-visit duration is abnormally long/short, outputs a urinary-disease (cystitis, urinary obstruction, kidney disease) early-warning alert. Especially useful for individualized health management in multi-cat households. Application: multi-cat homes,
This code automatically creates or logs in a remote user by sending openId/mobile/source fields to a remote endpoint when local tokens are absent. Performing identity creation and transmission without explicit disclosure or opt-in is dangerous because it can create external accounts, expose identifiers, and tie local activity to remote services outside the skill's declared purpose.
The skill declares broad operational behaviors that imply shell, network, environment, and file access, but it does not declare any explicit tool scope or permissions boundary. This increases the chance that an agent runtime grants more capability than users or reviewers expect, enabling unintended command execution, network access, or file handling.
The skill's documented scope expands beyond simple analysis into cloud history retrieval and report-link output, which introduces extra data access and disclosure pathways not central to the stated purpose. In a pet-health context, this can expose prior reports or linked resources tied to a user identity without clear expectation or consent.
The skill accepts local files and remote URLs and appears to send data to cloud APIs, yet it does not clearly warn users that uploaded videos and report queries may be transmitted off-device. This is risky because litter-box-area video can reveal household interiors, routines, and pet ownership details, creating privacy and compliance concerns.
The method returns a fixed Chinese-language heading string, which imposes a specific language on users. The file does not indicate any user choice, locale negotiation, or documented justification for restricting output to Chinese.
The script exposes a hidden --list mode that retrieves historical analysis reports via skill.get_output_analysis_list(open_id=...), which is outside the declared purpose of analyzing a supplied video or URL. Because this capability is not reflected in the skill description and is tied to internal identity state, it creates an unauthorized data-access surface that could expose prior user reports if invoked by an unintended caller or integrated agent flow.
This code defines the command description, argument help text, and runtime status/error messages in Chinese only. That imposes a specific language on all users without opt-in or any documented region-specific justification, which matches the language/locale policy violation criteria.
The call to OpenIdUtil.resolve_current_open_id(args.open_id, use_current=bool(args.open_id)) establishes internal identity context, and the script then uses that context to retrieve report history. In a skill whose stated function is one-shot video analysis, hidden identity resolution plus data retrieval is dangerous because it can enable cross-user data exposure, account confusion, or unauthorized access to stored behavioral/health-related records.
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
For local inputs, the code reads the entire file into memory and passes it as files to self.analysis(...), which transmits user-provided file data to an external service. While the code validates file existence and size, there is no confirmation prompt or explicit user-facing disclosure here that the file contents will be uploaded for remote analysis.
The skill implements a report-listing capability that goes beyond the stated trigger-and-analyze behavior in the manifest. This expands the accessible data surface and can expose prior analysis records or metadata to users who only expected single-input analysis, creating an authorization and privacy risk if report access is not tightly scoped.
The function forwards a local file path or remote URL into skill.get_output_analysis(...), which appears to invoke an external analysis backend, but the script provides no explicit disclosure, consent prompt, or boundary on what data may leave the local environment. In this skill context, videos of litter-box activity can reveal household interiors, pet ownership patterns, timestamps, and potentially humans or other sensitive background content, so silent off-system transmission creates a real privacy and data-handling risk.
Detected: suspicious.install_untrusted_source