Security audit

Brainstorming

Security checks across malware telemetry and agentic risk

Overview

This brainstorming skill is mostly coherent, but it can make durable repository changes, including writing a design file and committing it to git, without a clear separate approval step.

Install only if you want brainstorming sessions to produce repository artifacts. Before using it, tell the agent to ask before creating files, show the planned path and content or diff, and get separate approval before any git commit or worktree setup.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The skill goes beyond brainstorming by instructing the agent to commit a design document to git. That grants repository-modifying behavior in a skill whose stated purpose is design exploration, which can cause unintended state changes, unauthorized commits, or commit of sensitive/internal planning material without explicit user approval.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The instruction to create git worktrees sets up an implementation workspace, which exceeds the advertised brainstorming scope and initiates environment-changing actions. Even if intended as a convenience, it can trigger repository changes and workflow side effects without the user realizing the brainstorming skill can perform setup operations.

Description-Behavior Mismatch

Medium

Confidence: 96% confidence
Finding: The manifest and description present this as a pre-implementation brainstorming skill, but the body expands into writing files, committing to git, and initiating implementation setup. This scope mismatch is dangerous because users or calling systems may grant the skill access assuming low-risk ideation, while it actually contains instructions for persistent repository modification.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The requirement that the skill 'MUST use this before any creative work' is overly broad and can cause the skill to trigger for a large fraction of development tasks. Because the skill also contains file-writing and repo-modifying instructions later on, broad invocation increases the chance those side effects are reached in routine workflows where the user did not intend them.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill instructs writing and committing files without warning the user that repository state will be modified. Lack of disclosure is risky because users may invoke a brainstorming skill expecting conversational planning only, yet the agent may persist artifacts and create commits that affect audit trails, CI, or shared branches.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.