ClawHub
Back to skill

Security audit

Openclaw Intune Skill

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Intune and Entra administration skill with real tenant-admin power, but the sensitive behavior is purpose-aligned and guarded by confirmations, endpoint allowlists, and read-only mode.

Install only on machines and agent accounts trusted to administer the tenant. Prefer read-only Graph permissions plus INTUNE_READ_ONLY=true for reporting use, grant only the Graph permissions needed, protect the local user account/cache directory, and review confirmations carefully before device wipe/retire, policy, Conditional Access, group, app, or update-ring changes.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
This documentation expands the skill from Intune device management into broader Entra ID directory administration by exposing user, group, and membership-management operations. That scope expansion increases blast radius because a user invoking an ostensibly device-management skill could modify group membership, which can indirectly change app assignments, compliance targeting, and Conditional Access outcomes across the tenant.

Description-Behavior Mismatch

Low
Confidence
80% confidence
Finding
The documented ability to create Terms & Conditions and notification templates extends beyond a narrow device-management remit into broader tenant messaging and governance functions. While not inherently malicious, this creates scope drift and can enable unintended administrative changes or user-facing communications through a skill that may be trusted for a different purpose.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The invocation text is very broad and includes generic compliance, configuration, device, and policy terms, which could cause the skill to activate for loosely related enterprise-administration requests. In this skill, over-invocation is risky because the tool can perform real Microsoft Graph read and write operations against Intune/Entra resources, increasing the chance of unintended data access or administrative actions in the wrong context.

Credential Access

High
Category
Privilege Escalation
Content
#!/usr/bin/env bash
# get_token.sh — obtain (and cache) a Microsoft Graph app-only access token.
#
# Usage:
#   scripts/get_token.sh            # ensure a valid token is cached; print cache path
Confidence
82% confidence
Finding
access token

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.