Security audit

中国居民户口本识别与抽取（ADP）

Security checks across malware telemetry and agentic risk

Overview

This appears to be a legitimate household-register extraction skill, but it needs review because it sends highly sensitive identity documents to a cloud service and includes broader CLI powers than the narrow skill name suggests.

Install only if you are authorized to upload household-register documents to Laiye ADP and have reviewed the provider's privacy, retention, billing, and data-residency terms. Prefer npm or reviewed release artifacts over pipe-to-shell installers, keep API keys protected, restrict agents to the household-register OOTB extraction and query commands, and avoid the documented custom-app management commands unless you intentionally need administrative ADP control.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The README presents a general-purpose ADP CLI with broad parsing, extraction, task management, and app-management capabilities that materially exceed the declared skill scope of household-register recognition. This scope mismatch is dangerous because it can mislead agents or users into invoking unrelated high-privilege functionality, increasing the attack surface and enabling unintended document processing workflows beyond the advertised purpose.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: Documenting custom application creation, update, deletion, and AI-generated field configuration exposes functionality unrelated to a fixed household-register extraction skill. In an agent context, this can enable capability expansion, configuration tampering, or destructive actions that are unnecessary for the intended task and could be abused if the skill is trusted as narrowly scoped.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The README presents a broad general-purpose ADP CLI with parsing, extraction, tasking, and app-management capabilities rather than a narrowly scoped household-register extraction skill. This scope mismatch is dangerous because users or agents may invoke far more powerful functionality than the skill metadata implies, violating least privilege and increasing the chance of unintended data access or unsafe operations.

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The documented command set includes application-management and custom-app lifecycle operations that exceed the stated extraction-only purpose of the skill. In an agent context, this can enable unexpected administrative actions such as creating, modifying, or deleting extraction apps, increasing blast radius beyond simple document recognition.

Context-Inappropriate Capability

Medium

Confidence: 91% confidence
Finding: Custom app creation, update, deletion, and AI field-generation are not justified by the skill's stated household-register extraction function. These capabilities materially expand what an agent can configure or destroy, creating an unnecessary control plane that could be misused to alter extraction behavior, exfiltrate data through custom schemas, or disrupt existing applications.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The README instructs users to configure API credentials and use a cloud API for document parsing/extraction, but it does not explicitly warn that uploaded files and their extracted contents are sent to a remote service. Because this skill handles highly sensitive identity documents such as household registers, the missing disclosure increases privacy, compliance, and data-handling risk.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The README recommends fetching and executing remote installation scripts directly without warning users about integrity or trust risks. This is dangerous because any compromise of the remote source, repository, transport, or referenced script immediately becomes arbitrary code execution on the user's machine.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill instructs users to upload Chinese household-register images and extract highly sensitive personal data, including ID numbers, addresses, religion, and family relationship details, to a third-party remote ADP service, but it does not provide any explicit privacy, consent, retention, or cross-border data-handling warning. In this context, the omission is security-relevant because users may transmit regulated identity data without understanding where it is stored, processed, logged, or who can access it.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.