ClawHub
Back to skill

Security audit

iaiops-renewables

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a documented industrial monitoring skill, but it claims to be read-only while listing publish, push, and export tools that could affect telemetry and historian data.

Review this skill as a Review item before installing in passive-only or change-controlled OT environments. Confirm whether the runtime actually disables publish, push, export, and record-changing tools, and only use it where telemetry publishing and historian/UNS writes are explicitly allowed.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest and description repeatedly assert that the renewables edition is read-only, but the documented tool list includes mutating capabilities such as mqtt_publish, historian_push, export_data, stream_publish, uns_publish, and stream_publish_event. This mismatch can cause an operator or higher-level agent to trust the skill in environments where write actions are prohibited, enabling unintended data injection or exfiltration through industrial messaging and historian paths.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The workflow section states the edition surface is entirely read-only, yet the same skill exposes command surfaces that can publish or push data. In an OT/SCADA context, even 'only telemetry publishing' can be security-relevant because it may alter downstream systems, pollute UNS/MQTT topics, contaminate historians, or bypass change-control assumptions built around a supposedly passive skill.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.