Lark

Security checks across malware telemetry and agentic risk

Overview

This appears to be a legitimate Lark integration, but it gives an agent broad access to sensitive workplace data and destructive actions without enough safety guidance.

Review before installing. Use the narrowest Lark app scopes possible, avoid connecting an admin account unless required, and require explicit user confirmation before deleting content, changing permissions, removing people, approving or rejecting workflows, ending meetings, or reading and sharing transcripts or recordings.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill documents destructive actions like deleting messages without any requirement to confirm intent, describe consequences, or suggest guardrails. In an agent setting, this increases the chance of accidental or prompt-induced destructive changes to user data or communications.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill exposes broad capabilities to read messages, documents, calendars, contacts, meeting transcripts, recordings, approvals, and other potentially sensitive enterprise data, but it does not include privacy, consent, or least-privilege guidance. In an agent workflow, this can normalize bulk access to confidential business and personal information without adequate user awareness or restrictions.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal