ClawHub
Back to skill

Security audit

AuditClaw Aws

Security checks across malware telemetry and agentic risk

Overview

This skill appears to be a disclosed AWS compliance checker that uses scoped AWS read APIs and stores results locally, with some documentation and output-sensitivity caveats.

Install only if you are comfortable granting the included least-privilege AWS policy and writing findings into the local AuditClaw GRC database. Use a dedicated audit role or short-lived credentials, avoid root or broad managed policies, and treat connection-test output as sensitive because it includes account identity. Do not rely on the current KMS check as proof that key policies were reviewed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill advertises executable behavior and includes shell-invocable setup and run commands, but it does not declare explicit permissions for shell/code execution. This weakens the security boundary because users and platforms cannot accurately assess what capabilities the skill needs before invocation, increasing the chance of unexpected command execution in a sensitive environment with AWS credentials present.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding
The skill is presented as performing 15 read-only AWS checks, but the documented behavior includes local database writes and at least one AWS action, IAM credential report generation, that is not strictly read-only. This mismatch can mislead operators into granting trust or permissions under false assumptions, especially in environments handling cloud credentials and compliance evidence.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The module advertises both key rotation and key policy checks, but the implementation only evaluates rotation. In a compliance evidence collection skill, this can create a false sense of coverage and cause auditors or downstream automation to miss overly permissive or unsafe KMS key policies, weakening detection of real security issues.

Missing User Warnings

Medium
Confidence
80% confidence
Finding
The script prints AWS account_id and identity_arn directly to stdout during connection testing without any masking or warning. In CI logs, shared terminals, or captured agent transcripts, this can leak cloud account identifiers and role/user naming details that aid reconnaissance and environment mapping.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal