Security audit

LessonLoop

Security checks across malware telemetry and agentic risk

Overview

LessonLoop is a disclosed local memory helper for saving user feedback, with the main risk that corrections and preferences can be persisted for future sessions.

Install this only if you want the agent to save feedback as local memory. Avoid including secrets or sensitive personal details in lessons, configure the workspace path before using the scripts, and periodically review or delete entries in memory files and LessonLoop logs.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 89% confidence
Finding: The skill directs the agent to write to memory files and references helper scripts that append to persistent storage, yet it declares no permissions. This creates a capability-transparency problem: operators and users may not realize the skill can modify stored state, making unauthorized or unexpected persistence of user instructions more likely.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 78% confidence
Finding: The declared description presents a lightweight feedback-capture skill, but the referenced behavior includes escalation detection and activity reporting over prior logs, which expands scope beyond simple lesson storage. That mismatch can mislead reviewers about data processing and decision-making, increasing the risk of undisclosed monitoring, profiling, or routing changes based on stored history.

Vague Triggers

Medium

Confidence: 83% confidence
Finding: The README defines broad triggers like user corrections, new operating rules, repeated mistakes, and explicit 'remember this' instructions without clearly constraining consent, scope, or confirmation. In an agent system, this can lead to unintended activation and persistence of low-quality or sensitive lessons, especially when ordinary conversational feedback is interpreted as a durable rule.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The README explicitly describes scripts that write lessons to daily memory and log events, but it does not warn users that their feedback may be persisted or recorded. This creates a real privacy and transparency risk because users may provide corrective feedback assuming it is ephemeral, while the skill stores it for future use or evaluation.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill explicitly instructs writing corrections and rules into persistent memory files without any user-facing warning or confirmation at the point of modification. Because these memories can shape future sessions, silent persistence can store sensitive preferences, mistakes, or behavioral constraints beyond what the user expected.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.