Security audit

Elite Longterm Memory Local

Security checks across malware telemetry and agentic risk

Overview

This is a local memory skill, but it normalizes durable storage of conversation details and includes an unexplained callback URL despite privacy-first claims.

Install only if you want the agent to keep durable local memories across sessions. Avoid using it for secrets, credentials, regulated data, or sensitive personal/business details unless you first configure strict retention and review practices. Treat callback_url as a potential network egress feature and leave it unused unless you understand exactly what will be sent and where.

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: 84% confidence
Finding: The skill advertises broad automatic handling of local vector search and memory behaviors without clearly constraining when it should activate or what inputs should be excluded. In an agent environment, vague triggering increases the chance of unintended execution on sensitive conversations or unrelated tasks, leading to accidental persistence or command use.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: These sections instruct the agent to write conversation details into SESSION-STATE.md, MEMORY.md, daily logs, and a vector database, and to later compact, deduplicate, clean up, or back up stored data, but they do not present a clear user-facing warning or consent model. This is dangerous because users may disclose sensitive information without realizing it will be durably stored and further processed across multiple local stores.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The documented callback_url parameter introduces a network egress path despite the skill's 'local/privacy-first' positioning, yet no privacy, authenticity, or data-transfer warning is provided. If used carelessly, it can send processing status or associated data to external endpoints, undermining user expectations that all data remains local.

Ssd 3

Medium

Confidence: 95% confidence
Finding: The skill explicitly tells the agent to persist user-provided details into hot memory before responding, creating a default retention path for conversational content. This raises confidentiality and privacy risk because sensitive personal, business, or secret information may be stored indefinitely and later recalled in unrelated contexts.

Ssd 3

Medium

Confidence: 94% confidence
Finding: The workflow normalizes storing user details and decisions across session state, long-term memory, and logs as part of ordinary operation. In the context of a memory skill, that makes the issue more dangerous because the entire purpose is broad retention and recall, so accidental capture of sensitive information is likely rather than incidental.

VirusTotal

52/52 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.