Security audit

Neural Context Engine Free

Security checks across malware telemetry and agentic risk

Overview

This memory skill is mostly coherent, but it asks the agent to persist and automatically reuse conversation memory with weak user control and some overbroad activation language.

Install only if you are comfortable with the agent keeping a persistent local memory of conversation-derived facts, decisions, errors, preferences, and todos. Review the local database location, avoid storing secrets or sensitive personal data, and enable any external embedding provider only if you accept that relevant content may leave the local environment.

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 (5)

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The skill instructs the agent to automatically inject recent memory at session start and to store decisions, errors, and preferences during normal conversation. This broad activation logic can cause routine interactions to trigger storage and recall of user data without explicit user intent, increasing privacy leakage and over-collection risk.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The coverage scope is defined with very broad and ambiguous keywords, making it easy for unrelated conversations to activate the skill unexpectedly. In a memory-oriented skill, overbroad activation can lead to unnecessary retrieval or storage of potentially sensitive context.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill describes local storage and optional external embedding providers, but it does not present a prominent upfront warning that conversation content may be persistently stored and could be exposed to third-party services if optional integrations are enabled. Users may unknowingly provide sensitive data under the assumption that interactions are ephemeral and fully local.

Ssd 3

Medium

Confidence: 91% confidence
Finding: Automatically injecting recent memory at every session start violates data minimization principles and can surface previously stored user information even when it is not needed for the current task. This increases the chance of inadvertent disclosure of sensitive past context into new sessions, prompts, or outputs.

Ssd 3

Medium

Confidence: 90% confidence
Finding: The skill encourages persistent storage of user preferences, errors, and other conversational context without defining sensitivity boundaries, retention limits beyond some categories, or exclusions for secrets and personal data. In practice, this can accumulate sensitive information and make later unintended disclosure more likely.

VirusTotal

56/56 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.