Security audit

Ontology

Security checks across malware telemetry and agentic risk

Overview

This is a local Markdown-only ontology memory skill whose file writes are purpose-aligned and disclosed, though its broad triggers could cause unintended memory use.

Install only if you want the agent to maintain local structured memory in the workspace. Use explicit requests for writes, avoid storing secrets or raw credentials, and periodically review memory/ontology/graph.jsonl and memory/ontology/schema.yaml.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The skill advertises very broad trigger keywords such as 'graph', 'knowledge', 'typed', 'agent', and 'structured', which are common terms that may appear in many unrelated conversations. This can cause unintended activation of a skill that reads and writes persistent workspace memory, increasing the chance of unauthorized or surprising state changes across tasks.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The usage triggers are phrased as natural-language situations like 'Remember that...' and 'What do I know about X?', which are ambiguous and likely to occur during ordinary agent interaction. In this skill's context, ambiguous activation is more dangerous because the skill is designed to create, update, and link persistent entities in local files, so accidental invocation can silently alter memory or expose prior stored information.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.