Short Drama Writer

• Outcome: Pass
• Latest audit: May 29, 2026, 8:19 PM
• Version:
  2.3.6

Overview

This is a local writing helper, with the main caveat that one bundled script saves command inputs to a local history file.

Install only if you are comfortable with a local CLI writing tool. Avoid entering confidential story ideas, client material, secrets, or unreleased plans into scripts/script.sh unless you are comfortable with them being saved in a local history.log file under the skill's data directory.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: The script creates a persistent data directory and stores user activity in local log files even though its stated purpose is only content creation assistance. This introduces undisclosed data retention of potentially sensitive prompts, topics, or proprietary drafting ideas, which can expose user data to other local users, backups, or later collection by unrelated processes.

Context-Inappropriate Capability

Medium

• Confidence: 97% confidence
• Finding: Nearly every command routes user-supplied topics or prompts into a history file, creating systematic persistence of input that users would not reasonably expect from a simple CLI writing helper. Because prompts may contain confidential business plans, campaign ideas, or personal data, broad logging materially increases privacy and data exposure risk.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The helper function writes user input to a persistent history log without any user-facing notice, consent, or indication in normal command output. Silent collection is dangerous because users may enter sensitive text under the assumption the tool is ephemeral, leading to unintended local disclosure and compliance issues.

Ssd 3

Medium

• Confidence: 95% confidence
• Finding: User-provided content is recorded in plain text in a history file, making the stored data easy to read if the file is accessed by another local user, malware, support tooling, or backup systems. Plain-language logging of prompts is especially risky for a writing assistant because the content may include unpublished, personal, or commercially sensitive material.

Ssd 3

Medium

• Confidence: 97% confidence
• Finding: The command set consistently captures and retains all supplied user topics across multiple workflows, turning a basic assistant into a broad collection point for user-generated content. The skill context makes this more concerning because content-creation prompts often contain unpublished marketing plans, client data, or proprietary ideas unrelated to any legitimate need for persistent storage.

VirusTotal

63/63 vendors flagged this skill as clean.