Security audit

Memory Bridge

Security checks across malware telemetry and agentic risk

Overview

This is a coherent local memory skill, but it automatically stores and resurfaces user/session information with limited privacy boundaries or user-control safeguards.

Install only if you intentionally want local cross-session memory. Treat the created files as persistent private notes, avoid storing secrets or sensitive personal/business data, review them periodically, and delete MEMORY.md, memory/, and .learnings/ if you want to clear the memory.

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 (8)

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The guide promotes persistent storage of conversation context and automatic loading of memory files across sessions, but it does not disclose privacy, retention, or sensitive-data handling implications. In an agent skill, this can lead users to unknowingly store credentials, personal data, or sensitive work context on disk where later sessions will automatically re-ingest it, increasing exposure and accidental disclosure risk.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger phrases are broad, conversational, and overlap with normal user requests about memory or setup. That creates a real risk of unintended activation, which is more dangerous here because activation leads to persistent file reads/writes and cross-session retention of user information.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The skill says it will start automatically 'on wake-up' and detect or initialize memory without a clearly bounded trigger or consent boundary. In this context, ambiguous auto-activation is risky because it can cause unsolicited access to persistent files and automatic disclosure of retained user context in new sessions.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The description promotes automatic startup and memory recall but does not give a clear upfront warning that the skill will automatically read and write persistent files containing user-related data. Because the skill is specifically designed to retain identity, preferences, tasks, and corrections across sessions, the missing disclosure materially increases privacy and consent risk.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The setup flow directs the agent to create multiple files and directories in the workspace, but the user is not clearly warned beforehand that filesystem state will be modified. Silent or insufficiently disclosed file creation is a real safety issue because it changes the user's environment and establishes long-lived storage for potentially sensitive information.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The template explicitly instructs the agent to record important user information before responding, but it provides no user-facing notice, retention limits, consent mechanism, or privacy boundaries. In a memory or session-state context, this can lead to unnecessary collection and persistence of potentially sensitive user data, increasing the risk of privacy harm, over-retention, or later misuse.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The skill explicitly instructs the AI to persist and automatically surface user-specific information across sessions, creating a durable natural-language retention channel. This is dangerous because a later session, shared environment, or mistaken activation can disclose identity, preferences, and task context without fresh user intent or context-appropriate authorization.

Ssd 3

High

Confidence: 99% confidence
Finding: The core workflow tells the AI to collect, store, and reuse user identity, preferences, tasks, corrections, and related entities in persistent files. In skill context, this is a true privacy and confidentiality risk because it normalizes broad, ongoing accumulation of user data and instructs the agent to proactively reveal that remembered context in future sessions.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.