Security audit

Stuff Skill - 秩

Security checks across malware telemetry and agentic risk

Overview

The skill appears to build personal data apps, but its supplied scan evidence indicates it ships cloud API templates that could expose user data publicly without authentication.

Review and harden any generated cloud backend before using it with real personal data. Add authentication and authorization, restrict CORS to trusted origins, protect destructive operations, and confirm where data is stored and retained before deploying.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: The API enables cross-origin access from any website via Access-Control-Allow-Origin: * and exposes read/write/delete operations without any visible authentication or origin restriction. This makes the generated backend broadly callable by arbitrary web pages, increasing the risk of unauthorized data access or destructive actions if deployed as-is.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: The sample `functions/api/items.js` exposes GET, POST, and DELETE operations to anyone on the internet with no authentication, authorization, CSRF protection, or origin restrictions. An attacker could read, modify, or delete all stored application data remotely, and `Access-Control-Allow-Origin: *` further broadens abuse from arbitrary web origins.

Missing User Warnings

Low

Confidence: 92% confidence
Finding: The skill encourages cloud deployment and cross-device sync using third-party hosting and databases, but it does not give a clear privacy warning about what data may leave the device, where it will be stored, and what risks come with that choice. Users could unknowingly place personal inventory or tracking data into external services without understanding retention, access, or jurisdiction implications.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The trigger phrase set includes very generic requests such as '生成一个个人数据管理应用', which can overlap with ordinary user intents and cause the skill to activate in situations the user did not explicitly intend. That increases the chance of accidental invocation, context hijacking, or unreviewed generation of app artifacts based on sensitive personal data descriptions.

Natural-Language Policy Violations

Medium

Confidence: 82% confidence
Finding: The interface prominently presents Chinese-language display text, descriptions, triggers, and contract content without any indication of locale negotiation or user preference handling. This can mislead users, reduce informed consent, and cause incorrect activation or misunderstood outputs when the surrounding environment expects another language.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The user-facing guidance encourages cloud deployment and synchronization but does not warn users that their app data will be stored on Cloudflare-managed infrastructure and transmitted over the network. This creates a transparency and privacy risk because users may unknowingly place sensitive personal data into third-party hosted storage.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.