Security audit
Security checks for vulnerabilities and agentic risk
This appears to be a real cloud object-detection skill, but it sends sensitive surveillance media to remote services while silently creating or reusing identities and caching tokens locally.
Install only after confirming you are comfortable sending images, videos, URLs, identity metadata, and report requests to the configured lifeemergence services. Use a dedicated workspace or account, review data retention and token storage expectations, and avoid using sensitive surveillance footage unless the publisher provides clear consent, privacy, deletion, and authorization controls.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The described use of external APIs, authentication, remote account creation/login, local file creation, environment discovery, and identity persistence substantially exceeds the expectations created by a simple vision-analysis label. In this context, surveillance media may contain sensitive personal data, so undisclosed transfer, authentication, and storage behavior is especially risky.
The YARA hit indicates suspicious manifest/metadata characteristics consistent with tool or metadata poisoning, and the document also contains odd mixed-purpose wording and corrupted text near the header. While not proof of maliciousness by itself, in combination with the broader description-behavior mismatches it suggests the manifest may be crafted to mislead hosts or obscure true behavior.
---
name: "basic-object-detection-analysis"
description: "Detects people, vehicles, non-motorized vehicles, pets, and parcels appearing in the target area. Supports video stream and image detection, suitable for general security surveillance scenarios. | 基础目标检测技能，检测出目标区域内出现的人、车、非机动车、宠物、包裹，支持视频流和图片检测，适用于通用安防监控场景"
version: "1.0.20"
license: "MIT-0"
---
# 🎯 Basic Object Detection Skill | 基础目标检测技能
> **智能健康/识别分析中枢** · 图片/视频智能分析 · 结构化报告 · 历史报告云端查�
The skill does not clearly warn users that uploaded files and provided URLs are sent to a cloud/API service for processing. Because surveillance images and video can contain sensitive personal or facility data, undisclosed remote transmission creates significant privacy, compliance, and data-handling risk.
The User model stores sensitive identity and authentication material including username, real name, email, token, and open_token, yet the skill description is only about object detection. Unnecessary storage of authentication tokens is high risk because compromise of the local SQLite database could expose credentials or enable account takeover in systems outside the skill.
The skill includes logic to derive or create a persistent user identity, read a workspace API-key-derived identity file, and create local user records unrelated to basic object detection. In the context of a surveillance/object-detection skill, this is dangerous scope expansion because it silently establishes account identity and persistence that can be reused for external service access without clear user consent.
The HTTP helper automatically logs in or registers users against external services, populates tokens, caches them locally, and retries authorization transparently. For a skill advertised as basic object detection, this hidden authentication and outbound account activity materially exceeds expected behavior and can cause undisclosed data transmission and account creation.
The skill declares broad operational behavior that implies shell execution, network access, local file handling, and environment use, but it does not declare any tool scope or permission boundaries in the manifest. In an agent setting, missing explicit capability constraints increases the risk of over-privileged execution and makes it harder for hosts to sandbox or review what the skill is allowed to do.
The skill documentation broadens functionality from immediate object detection to cloud history lookup and report-link retrieval, which expands accessible data beyond the submitted media. In a surveillance setting, historical reports can reveal sensitive operational patterns and prior detections, making undocumented expansion materially risky.
The history-report trigger phrases are broad enough to activate cloud report queries from ordinary conversational requests, increasing the chance of unintended retrieval of sensitive historical data. In an agent context, overly permissive intent triggers can cause data access without a sufficiently deliberate user action.
The documentation says attachments are automatically saved as local files but does not present this as a clear user warning. Automatic local persistence of uploaded surveillance media increases the risk of unintended retention, later exposure, and mishandling on shared systems.
Automatic internal identity initialization, reuse, and silent association of reports are not justified by the stated object-detection function and create hidden identity linkage. This can enable tracking across sessions and unauthorized access to prior reports if local defaults or reused identities are applied incorrectly.
The API documentation describes a pet health analysis service while the skill metadata claims the skill performs basic object detection for security surveillance. This mismatch can mislead reviewers and integrators about the skill’s actual functionality, causing the wrong backend service, data types, or data handling expectations to be trusted and increasing the risk of unintended data exposure or misuse.
The skill includes a history/listing capability via show_analyze_list(open_id) that is outside the stated purpose of basic object detection. In a surveillance context, exposing prior analysis jobs or user-associated history can reveal sensitive metadata and prior media-processing activity, especially if access control around open_id is weak or implicit.
The skill sends image/video input to a remote analysis service through skill.get_output_analysis(...) without any explicit notice, consent, or privacy warning to the user. Because the skill is designed for security surveillance media, this can expose highly sensitive footage, personal data, and location/activity information to external systems unexpectedly.
Detected: suspicious.install_untrusted_source