Speak 69378f

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only learning workflow with no code or privileged access, though its activation wording is broad and may cause accidental use.

Safe to install from a security standpoint. Expect possible accidental activation on ordinary mentions of "speak" or SpeakNow; install only if you want this video-derived Speak workflow available, and consider narrowing triggers if you maintain it.

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 trigger field includes very broad phrases like "speak" and "Speak", which are common natural-language terms and likely to appear in unrelated conversations. This can cause the skill to activate unexpectedly, creating prompt-routing confusion and making it easier for irrelevant or low-quality content to override the user’s actual intent.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The activation scenarios are ambiguous because they treat generic mentions like "用户说 'speak'" or "用户需要speak相关帮助" as sufficient to invoke the skill. In context, this skill is a loosely assembled video-derived workflow with merged unrelated sources, so ambiguous activation increases the chance of accidental invocation and incorrect task handling.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal