Security audit
Security checks for vulnerabilities and agentic risk
The skill appears to provide cloud-based pet media analysis, but it also silently links users to accounts, stores tokens locally, and exposes cloud history/report behavior with limited user control.
Review this before installing if you are comfortable with pet videos or video URLs being sent to lifeemergence.com/open.lifeemergence.com services, with an internal user identity being automatically created or reused, and with local storage of user/token records for cloud report history. Avoid using sensitive household footage unless the cloud processing, retention, and account-linking model is acceptable.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
Outbound HTTP requests, auth token handling, workspace file I/O, environment/context detection, identity resolution, and DAO/database persistence are all broad infrastructure capabilities that exceed the declared pet-behavior purpose. In context, this is dangerous because the skill processes user media and report history, so hidden identity linkage and storage materially increase privacy and data-governance risk.
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
---
name: "pet-behavior-detection-analysis"
description: "Identifies common abnormal pet behaviors such as scratching, biting, destructive chewing, jumping, digging, chasing, and separation anxiety, helping owners understand their pet's habits. | 宠物行为识别技能，识别抓挠、啃咬、拆家、跳跃、刨地、追逐、独处焦虑等常见宠物异常行为，帮助主人了解宠物行为习惯"
version: "1.0.22"
license: "MIT-0"
---
# 🐾 Pet Behavior Recognition Skill | 宠物行为识别技能
> **智能分析中枢** · 图片/视频智能分析 · 结构化报告 · 历史报告云端查询
---
## 🧭 技能概�
This file exposes generic CRUD helpers and arbitrary HTTP wrappers that can send requests to caller-supplied URLs, which is far broader than the declared pet-behavior analysis purpose. In an agent-skill context, this expands the capability surface to unintended external actions, enabling data exfiltration, unauthorized API access, or use of the skill as a general network proxy if other components can influence the URL or payload.
The http_post/http_put/http_get/http_delete methods accept arbitrary URLs and pass through arguments directly, giving the skill broad outbound network access unrelated to pet-behavior detection. In this context, that is dangerous because the skill manifest does not justify acting as a generic HTTP client, so these methods could be abused for SSRF-like access, exfiltration, or calling attacker-controlled endpoints.
This utility performs authenticated API access, token handling, user lookup, token persistence, and automatic account bootstrap logic that is unrelated to pet behavior detection. In the context of a pet-analysis skill, this creates undisclosed identity and network side effects, enabling remote data transmission and account activity far beyond the declared functionality.
The helper can automatically register or log in a remote user by sending openId/mobile data to a backend endpoint with silent and register flags enabled. For a pet behavior detection skill, this is unjustified and dangerous because it allows undisclosed remote account creation and identity linkage without informed user action.
The skill advertises execution of local scripts, file handling, network access, environment use, and shell commands but does not declare any explicit tool scope or permission boundaries. In an agent environment, this broad undeclared capability increases the chance of over-privileged execution and makes it harder for policy enforcement layers to restrict dangerous actions.
The document first frames the capability as behavior recognition, then describes it as a health/disease diagnostic system that identifies health risks and disease symptoms. This inconsistency is dangerous because users may consent to routine behavior analysis without realizing they are using a medical-style diagnostic workflow with different sensitivity, liability, and data-handling implications.
The description does not clearly warn users up front that local files and remote URLs will be transmitted to a cloud API for analysis. Because the skill handles potentially sensitive user media and automatically associates identity internally, lack of prominent disclosure undermines informed consent and increases privacy risk.
The history-report trigger phrases are broad enough that ordinary conversation about reports may automatically invoke cloud history lookup. In this skill's context, that can unintentionally retrieve prior user-linked analysis records and expose historical metadata without a sufficiently explicit request.
The document describes pet health analysis APIs even though the skill is presented as pet behavior detection. This kind of scope mismatch can cause the agent or integrators to call unintended backend capabilities, exposing health-reporting functions, data flows, or permissions that were not reviewed for this skill’s declared purpose.
The listed operations include starting health analysis, retrieving health analysis results, querying historical reports, and exporting full reports, which indicates access to report-oriented health workflows rather than behavior detection. If wired into the skill, this could enable unauthorized collection or export of sensitive pet health data under a misleading skill label.
The function sends local file paths or remote URLs to a backend analysis service without an explicit user-facing warning that video content may be transmitted off-device. Video files can contain sensitive household, location, and personal information, so lack of disclosure undermines informed consent and privacy expectations.
The skill exposes user-specific history retrieval via an open_id-linked listing function, which is not necessary for one-off pet behavior analysis and broadens the data-access surface. In combination with hidden identity handling elsewhere in the script, this can enable unnecessary access to prior analysis records and creates privacy risk if identities are resolved or reused without clear consent.
The script can write results to an arbitrary local path via --output, which exceeds an analysis-only expectation and may overwrite files if invoked in an unsafe context. This is not inherently malicious, but hidden or undocumented filesystem write capability increases risk when the skill is integrated into larger agent workflows.
The script silently resolves an internal user identity through OpenIdUtil without clear disclosure, which can bind analysis actions and history to a user account unexpectedly. Hidden identity resolution is a privacy and trust issue, especially when paired with history retrieval functionality.
Detected: suspicious.install_untrusted_source