Openclaw Plugin Dev

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:32 PM UTC
• Version:
  1.0.0

Overview

This is a documentation-only OpenClaw plugin development guide, but its logging examples could expose sensitive LLM conversation data if copied without safeguards.

Reasonable to install as a guide. If you use its examples to build a plugin, do not log full prompts, system prompts, chat history, or outputs by default; add opt-in controls, redaction, restricted file permissions, and a retention or deletion policy.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill explicitly promotes logging all LLM API requests and responses, which can capture prompts, system prompts, chat history, secrets, and personal data, yet it provides no warning, consent guidance, redaction strategy, or retention controls. In a plugin-development context this is especially risky because it gives implementers a ready-made surveillance pattern that may be deployed broadly and silently.

Missing User Warnings

Medium

• Confidence: 86% confidence
• Finding: The early examples teach interception of `llm_input` and `llm_output` events and correlation of full request/response data, normalizing collection of potentially sensitive model inputs and outputs. Although framed as plugin mechanics, the absence of any caution about privacy, minimization, or safe handling makes the guidance unsafe by default and increases the chance of accidental data leakage.

VirusTotal

63/63 vendors flagged this skill as clean.