Munger Observer

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only reflection skill that reads daily memory and session logs for a disclosed mental-model review, with no evidence of exfiltration, hidden execution, or destructive behavior.

Install this only if you are comfortable with the agent using today's memory and session logs to produce a private thinking review. Avoid optional daily scheduling unless you want recurring reviews, and be mindful that sensitive information stored in memory or logs may be summarized in the output.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The trigger phrases are broad enough to match ordinary requests such as 'review my thinking' or 'check for blind spots,' which can cause the skill to activate in situations the user did not clearly intend. Because the skill then performs retrospective analysis over user activity, overbroad invocation increases the chance of unexpected processing of sensitive context and unintended autonomy.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly directs reading today's memory file and session logs, which may contain sensitive user data, but it provides no user notice, consent step, or scope limitation. This creates a privacy and data-minimization risk because the skill may inspect more activity data than the user expects for a simple 'mental models' review.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal