Lark

Security checks across malware telemetry and agentic risk

Overview

This Lark coordination skill is broad and can touch sensitive workplace data, but its behavior is disclosed, instruction-only, and guarded by mode and confirmation rules.

Install with a dedicated least-privilege Lark app. Keep Counselor Mode unless you intentionally want routine write actions, and review confirmations carefully for messages, approvals, spreadsheet edits, scheduling changes, and any senior-stakeholder or cross-functional communication.

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 (1)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The trigger list includes broad natural-language phrases such as 'meeting summary' and 'review pending approvals' that are plausible in ordinary conversation, making accidental or prompt-injected invocation more likely. In a skill tied to organizational coordination and potentially sensitive Lark actions, unintended activation could expose data, change workflow state, or steer the agent into acting in the wrong operational mode.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal