my skill

Security checks across malware telemetry and agentic risk

Overview

This is a low-risk, instruction-only skill for reviewing and improving OpenClaw SKILL.md files.

Safe to install for SKILL.md quality review. Avoid giving it paths to private or unrelated files, and review suggested edits before applying them.

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (3)

Medium

Confidence: 94% confidence
Finding: The manifest description indicates the skill may be triggered by a very generic description-only phrase, which can cause unintended activation in contexts unrelated to skill-quality auditing. Overly broad triggers increase the chance that the skill is invoked on arbitrary content, potentially chaining into automated review/fix workflows without sufficient user intent or scope validation.

Medium

Confidence: 84% confidence
Finding: The skill's invocation description is broad enough to match many generic requests about reviewing, optimizing, diagnosing, or checking a skill file, which can cause the agent to activate in contexts the user did not specifically intend. Overly broad auto-triggering increases the chance of misrouting user requests, applying the wrong workflow, or exposing unrelated skill content to this reviewer unnecessarily.

Medium

Confidence: 85% confidence
Finding: The guidance explicitly recommends very broad trigger phrases such as 'search' and 'find information,' which are common in ordinary conversation and can cause accidental skill activation. In a skill-routing system, overly broad triggers can hijack user requests from the general agent flow, causing inappropriate tool use, unexpected data access, or bypass of more context-appropriate handling.

65/65 vendors flagged this skill as clean.