Security audit

Intelligence Brain

Security checks for vulnerabilities and agentic risk

Overview

The skill appears purpose-aligned, but it handles internal company intelligence with automatic cloud model routing and scheduled data deletion without enough user control or retention safeguards.

Review this skill carefully before installing in a business environment. Use it only if you are comfortable with internal company data potentially being routed to external model providers, and configure local-only processing or approval gates for sensitive data. Confirm retention settings before enabling cleanup, especially where source records may be needed for audit, legal, incident, or business continuity reasons.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (4)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill explicitly describes a pipeline where source information may be archived or cleared within 24 hours and later reiterates automatic deletion/cleanup behavior, but it does not warn users about irreversible data loss, retention implications, or the need for confirmation before destructive actions. In an agent skill that processes company intelligence, this creates a real risk of unintended loss of source material, audit evidence, or business-critical context.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill states that every piece of incoming information is automatically assigned a classification level that determines which model it can be sent to, including private or cloud-hosted models, but it does not clearly warn users that sensitive company data may be transmitted outside the local environment. Because the skill is designed as a centralized company-intelligence engine, misclassification or over-broad automation could expose confidential business information to external providers.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The model-routing section explicitly permits internal company data classified as '内部' to be processed by paid cloud models and public data by free models, without an explicit warning about external transmission, vendor retention, cross-border processing, or misclassification risk. In context, this is more dangerous because the entire skill is built to ingest, synthesize, and route cross-department intelligence, so a single flawed label or policy could leak aggregated sensitive insights at scale.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The playbook explicitly instructs deletion of raw and parsed data on a schedule, but it does not require operator confirmation, backup verification, exceptions for legal/incident retention, or a user-facing warning about irreversible loss. In an agent skill, operational instructions can be executed automatically or normalized as safe behavior, making unintended data destruction materially more likely.

Static analysis

No suspicious patterns detected.