Agent Reflective Memory

Security checks across malware telemetry and agentic risk

Overview

This is a local agent-memory helper that stores user-provided experiences on disk, which matches its purpose but requires care with sensitive inputs.

Install only if you want a local memory log for agent activity. Do not store secrets, credentials, private customer data, regulated data, or confidential prompts unless you are comfortable with them remaining in memory_store.json and being returned by later searches; delete that file to clear stored memory.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (3)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly stores, reflects on, and retrieves past agent actions and decisions, which can include prompts, operational context, database activity, or other sensitive data. Because the description provides no warning about sensitive-data handling, retention, or reflection risks, users may enable it in contexts where secrets, personal data, or confidential business information are ingested and later exposed through summaries or queries.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The engine persists raw agent context, actions, results, and metadata to a local JSON file without any consent flow, minimization, redaction, or protection controls. In an agent setting, these fields can easily contain secrets, personal data, prompts, tokens, or internal task history, so writing them to disk creates a durable disclosure risk if the host is shared, backed up, logged, or later inspected.

Ssd 3

Medium

Confidence: 96% confidence
Finding: The skill stores full user-supplied context, action, result, and metadata verbatim and later returns matching records through query operations, which can expose sensitive inputs in plain language. Because this is a reflective memory component for autonomous agents, the stored data is especially likely to include confidential prompts, operational details, credentials, or user data that may be retrievable by later queries or consumers of the memory store.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal