Book Writing Assistant

Security checks across malware telemetry and agentic risk

Overview

This is a coherent book-writing helper whose main tradeoff is that it saves manuscript project details locally for continuity.

Install this only if you are comfortable with the agent saving book project details in a local book-data.json file for later use. Review or delete that file if you do not want old manuscript details, research notes, or continuity data influencing future sessions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs activation on broad, casual phrases such as 'I'm working on my book' and similar long-form writing mentions, which can cause the agent to invoke this skill in conversations that are only loosely related. Over-broad routing increases the chance of unintended data handling, irrelevant persistence to `book-data.json`, and context hijacking away from the user's actual task.

Missing User Warnings

Low
Confidence
84% confidence
Finding
The eval explicitly expects the skill to write to a local file (`book-data.json`) without any mention of user awareness, confirmation, or safeguards. In an agent context, silent persistence to local storage can create unintended state changes, overwrite existing data, or store sensitive user-provided manuscript details without consent, making this a real security and privacy concern even if the file operation itself is low risk.

Missing User Warnings

Low
Confidence
88% confidence
Finding
The eval explicitly expects the skill to write to book-data.json, which indicates persistent state modification without any mention of notifying the user or obtaining consent. In a writing assistant, silently saving research notes is plausible product behavior, but it still creates a transparency and consent issue because user content is being stored and may persist across sessions.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal