Security audit
Security checks for vulnerabilities and agentic risk
This skill is packaged as a business migration planner but actually steers agents toward helping retrieve private WeChat data from a protected iPhone, with broad automatic activation and no legal or consent guardrails.
Review this skill carefully before installing. It does not run code, but it is misbranded and can be invoked broadly to help with access to private communications on a protected phone. It should be rewritten to accurately describe its purpose and limited to lawful estate, Apple/Tencent support, backup, and professional legal/forensic channels, with explicit refusal of security-bypass or unauthorized data extraction help.
The README explicitly centers the skill on extracting WeChat data from a locked iPhone but provides no warning about consent, legality, privacy, or the risks of accessing protected personal data. For a user-facing skill, omission of these safeguards makes misuse more likely and normalizes assistance for unauthorized data access.
The skill's core requirement is framed around retrieving WeChat data from a protected iPhone, which directly concerns access to highly sensitive private communications. Because the data belongs to a deceased spouse and the device is blocked by anti-theft protections, the context is especially dangerous: the skill can facilitate unauthorized disclosure, circumvention attempts, or digital-estate abuse under a sympathetic pretext.
The trigger phrases convert a sensitive data-access objective into routine activation language, allowing users to invoke privacy-invasive help through innocent-sounding prompts. This lowers friction for harmful use and makes the risky capability easier to discover and operationalize at scale.
The file presents itself as a benign 'USA Business Migration Planner' while the actual documented purpose is assisting with obtaining WeChat data from a deceased person's iPhone blocked by Face ID/theft protection. This mismatch is dangerous because it disguises a sensitive device/data-access objective under unrelated branding, increasing the chance the skill is invoked, approved, or distributed without proper scrutiny.
The skill explicitly packages a sensitive objective—extracting WeChat data from a locked deceased person's iPhone—as a reusable, repeatable workflow. This is dangerous because it normalizes privacy-invasive access to personal communications and may facilitate unauthorized handling of protected device and app data under benign planning language.
The trigger phrases are broad enough to match common help requests while routing into a skill centered on sensitive device/data access. This expands accidental activation and increases the chance that ordinary user queries are steered toward unsafe or privacy-invasive assistance.
The trigger text semantically directs the skill to help obtain private WeChat data from a device, which is a data-exposure objective even without explicit exfiltration terminology. In this context, the skill makes a sensitive personal-data access goal easier to invoke and operationalize, increasing privacy, legal, and abuse risks.
The skill’s declared identity as a 'USA Business Migration Planner' is materially inconsistent with its actual payload, which centers on extracting WeChat data from a deceased person’s iPhone behind Face ID/theft protections. This mismatch is dangerous because it disguises a sensitive-device/data-access workflow under an unrelated benign label, increasing the chance of stealthy invocation, reduced scrutiny, and misuse for unauthorized access guidance.
The top-level description uses broad trigger terms like general-help, apple, iphone, and face, which can match many ordinary user requests unrelated to the stated scenario. This is risky because it allows a sensitive-access-oriented skill to activate too broadly, potentially steering benign conversations toward guidance on personal-data extraction from protected devices.
The description explicitly promotes help for extracting WeChat data from a deceased wife’s iPhone while mentioning known credentials and a Face ID/theft-protection barrier, but provides no authorization or legal-validation boundaries. That is dangerous because it frames access to highly sensitive personal communications on a protected device as a routine support task, enabling privacy invasion, unauthorized access, or guidance adjacent to security-bypass behavior.
The skill wraps a sensitive objective—recovering private WeChat data from a deceased person’s protected iPhone—in a benign workflow/planning narrative about reusable help and practical checklists. This context makes it more dangerous because it operationalizes and normalizes a high-risk access goal while obscuring the privacy and authorization issues, making abusive use easier at scale.
The keyword list contains highly ambiguous everyday terms such as general-help, apple, iphone, face, pro, and max without any scoping safeguards. In context, this makes accidental or covert triggering more likely and expands exposure of a skill that normalizes assistance around accessing data on a protected personal device.
The skill is mislabeled as a benign 'USA Business Migration Planner' while its actual content operationalizes access to WeChat data from a deceased spouse’s iPhone despite Stolen Device Protection. This deceptive packaging can evade review, misroute invocation, and conceal a sensitive device/data-access objective behind an unrelated title.
The activation criteria are ambiguous and broadly framed around generic help categories, making it unclear when this skill should be invoked. Because the skill concerns a sensitive data-retrieval objective, vague activation rules materially raise the risk of inappropriate or overbroad deployment.
The skill discusses extracting WeChat data from a deceased spouse’s iPhone but provides no privacy, consent, legal, or data-handling warnings. In this context, omission of safeguards normalizes access to highly sensitive personal communications and may facilitate unauthorized access or secondary privacy harms.
The skill frames a repeatable workflow for obtaining WeChat data from a protected iPhone as ordinary planning assistance, which normalizes a sensitive access objective and reduces friction for misuse. In the context of device protections and third-party personal data, this is dangerous because it can facilitate privacy invasion and evasion of access controls.
Broad trigger keywords like 'apple', 'iphone', 'face', 'pro', and 'max' can cause the skill to activate during unrelated conversations, unnecessarily surfacing a workflow centered on sensitive device/data access. Overbroad routing increases exposure and the likelihood of unsafe assistance being offered without clear user intent.
The trigger examples repeatedly instruct the agent to help achieve a locked-iPhone data-retrieval goal, turning a sensitive and potentially unauthorized access objective into a standardized request pattern. This directly increases usability for harmful intent by making invocation easy and routine.
The default prompt explicitly requests help for extracting WeChat data from a deceased spouse's iPhone that is blocked by Stolen Device Protection and Face ID requirements. That is guidance aimed at bypassing platform security controls and accessing highly sensitive third-party data, which could facilitate privacy violations, unauthorized access, or unlawful forensic extraction.
The prompt is oriented toward obtaining private data from another person’s device, and the fact that the subject is deceased does not eliminate the sensitivity or the need for formal authorization. In this skill context, the mismatch between the declared business-planning purpose and the actual device-data extraction request makes the behavior more dangerous because it disguises a privacy-invasive objective inside an innocuous-seeming skill.
The skill’s declared identity as a 'USA business migration planner' materially conflicts with its actual function: guidance about accessing data on a deceased person’s locked iPhone and extracting WeChat data. This kind of scope deception can bypass user expectations, policy routing, and review controls, increasing the chance the skill is invoked in inappropriate or privacy-invasive contexts.
The skill addresses extraction of WeChat data from a deceased person’s locked iPhone but provides no warning about legal authority, privacy obligations, or the sensitivity of credentials and personal communications. In context, that omission is dangerous because users may interpret the skill as endorsing access to highly sensitive data without verification of consent, estate authority, or platform rules.
Using 'Help me' as part of a trigger is dangerously overbroad because it overlaps with ordinary conversation and can cause accidental invocation. In this skill, unintended activation is more serious because the underlying topic involves sensitive device access and private-message extraction, so casual user requests could be routed into high-risk guidance without clear intent.
The trigger phrases are written as broad, natural-language requests that can activate the skill in ordinary conversations without clearly signaling that it handles highly sensitive device-access and message-extraction scenarios. In context, this increases the chance the agent will operationalize privacy-invasive assistance from benign-looking prompts.
Although the README describes a generic reusable workflow, the concrete use case is narrowly centered on overcoming iPhone access barriers to retrieve WeChat data. Framing a sensitive access goal as ordinary workflow assistance can normalize unauthorized data access and help evade simple policy checks that look for explicit bypass language.
