gstack CEO Review

• Outcome: Review
• Latest audit: May 28, 2026, 7:34 AM
• Version:
  1.0.0

Overview

This is an instruction-only plan-review skill, but it tells the agent to persist review summaries that may contain sensitive project details without asking the user first.

Install only if you are comfortable with the agent saving CEO review summaries for later use. Avoid using it on confidential roadmaps, unreleased architecture, security issues, or customer data unless you can disable or review the memory-writing behavior. The skill is not executable malware, but its default persistence deserves explicit user review.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Intent-Code Divergence

Medium

• Confidence: 95% confidence
• Finding: The skill claims it only reviews plans and makes no code changes, yet it also instructs saving the review summary to `memory/`, which is a persistent side effect. This mismatch can mislead users and operators about what the skill actually does, and may cause unreviewed retention of project details that were only intended for transient analysis.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger description uses broad, everyday phrases like 'review a plan,' 'challenge this,' 'poke holes,' and 'think bigger,' which increases the chance of accidental invocation in unrelated conversations. Unintended activation can inject strong behavioral constraints into normal user interactions and may also trigger the memory-persistence behavior on content the user did not intend to process with this skill.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The instruction to save a summary to `memory/` introduces persistent storage without warning the user that their plan contents, critiques, or sensitive project details may be retained. Silent persistence is risky because users may share confidential roadmap, architecture, or business information under the assumption that the interaction is ephemeral.

Ssd 3

Medium

• Confidence: 96% confidence
• Finding: Persisting CEO review summaries can retain sensitive user-provided information such as internal architecture, strategic plans, product gaps, and security concerns, which may later be surfaced out of context or to unintended parties. In this skill's context, the reviewed material is likely to be especially sensitive because it concerns roadmap, implementation alternatives, and threat modeling.

VirusTotal

VirusTotal findings are pending for this skill version.