Security audit

Workspace Hygiene Publish

Security checks across malware telemetry and agentic risk

Overview

This skill locally checks an OpenClaw workspace and writes a hygiene report, with optional memory-file merging, which matches its stated purpose.

Install only if you want an agent to inspect the selected workspace's memory, project folders, and documentation. Run it intentionally on the desired workspace, expect a hygiene report to be written under projects/system, and use --fix only when local memory-file consolidation is acceptable.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill clearly performs file reads and writes by scanning the workspace, auto-consolidating memory files, and writing reports, yet the manifest declares no permissions. That mismatch is dangerous because it hides the skill's actual capabilities from reviewers and users, reducing informed consent and weakening policy enforcement around file access.

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The script always writes a hygiene report into the target workspace, and may create the report directory as well, even though the tool is framed as an audit utility and exposes a 'report-only' mode. In an agent setting, unexpected writes can violate operator expectations, alter workspace state, and overwrite or create files in sensitive project areas during what should be a read-oriented assessment.

Intent-Code Divergence

Medium

Confidence: 97% confidence
Finding: The CLI help claims '--report-only' will avoid applying fixes, but the implementation still creates directories and writes a report file under the workspace. This mismatch is dangerous because users and higher-level agents may rely on the flag for non-mutating operation, leading to unintended filesystem changes and policy violations in environments where audits are expected to be side-effect free.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The trigger phrase "audit files" is broad enough to match many unrelated user requests, which can cause the skill to activate outside its intended hygiene context. Because this skill can write reports and auto-consolidate files, accidental invocation increases the chance of unintended workspace inspection or modification.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The documentation states that the skill writes a hygiene report and auto-consolidates timestamp-format memory files, but it does not prominently warn that running the skill may modify workspace contents. This lack of disclosure is risky because users may expect a read-only audit and unintentionally authorize changes to memory artifacts or project files.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.