A security-focused Openclaw Skills companion that assesses URLs for threats and aligns web navigation with user intent.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install preclick
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install preclick using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is PreClick URL Security Verifier?
PreClick is a sophisticated companion skill designed for AI agents to verify target web resources before any interaction occurs. It provides a robust security layer by checking URLs for phishing, malware, and suspicious patterns. Unlike traditional scanners, it includes intent-aware analysis to ensure that a destination truly matches what the user aims to accomplish, making it an essential addition to any developer's repertoire of Openclaw Skills.
By integrating this skill, agents can proactively avoid malicious sites and ensure that activities like credential submission or file downloads are performed on legitimate, safe domains. The system provides clear directives, enabling agents to act autonomously while maintaining high safety standards.
PreClick URL Security Verifier Use Cases
- Verifying links provided by users or discovered during browsing before navigation.
- Assessing safety before submitting credentials to a login page.
- Checking file download links for potential malware or threats.
- Ensuring navigation targets align with specific user goals like booking or purchasing.
- Following redirect chains to identify hidden security risks in complex web flows.
How PreClick URL Security Verifier Works
- The agent identifies a URL that requires navigation or interaction and determines if a specific user intent is associated with it.
- The skill triggers a scan using threat-focused or intent-aware analysis tools based on the available context.
- PreClick evaluates the target resource and returns an agent_access_directive such as ALLOW, DENY, or REQUIRE_CREDENTIALS.
- The system calculates a risk_score and checks intent_alignment against the user's stated purpose to identify potential mismatches.
- The agent acts on the directive, either proceeding with the task or warning the user based on the detailed scan results.
PreClick URL Security Verifier Setup
To get started with this skill, install the plugin and restart the Openclaw gateway using the following commands:
openclaw plugins install @cybrlab/preclick-openclaw
openclaw gateway restart
Verify the installation by checking the plugin and skill lists to ensure the environment is ready for Openclaw Skills integration:
openclaw plugins list | grep -i preclick
openclaw skills list | grep -i preclick
PreClick URL Security Verifier Data Schema & Taxonomy
The skill returns a structured payload containing several key metrics used for decision-making within Openclaw Skills:
| Field | Description |
|---|---|
agent_access_directive |
Primary action command (ALLOW, DENY, RETRY_LATER, REQUIRE_CREDENTIALS). |
risk_score |
A float (0.0 to 1.0) representing the threat probability. Lower is safer. |
confidence |
A float (0.0 to 1.0) indicating analysis certainty. |
intent_alignment |
Signal indicating if the page matches the user's goal (misaligned, no_mismatch_detected, inconclusive). |
analysis_complete |
Boolean indicating if the full analysis finished. |
agent_access_reason |
A text explanation of why a specific directive was given. |
PreClick URL Security Verifier Advanced Features
- Intent-aware verification that compares site content against specific user goals to prevent redirection to legitimate but irrelevant sites.
- Support for non-blocking asynchronous scans via compatibility tools for clients without native MCP task support.
- Native OpenClaw task-method proxy tools for comprehensive lifecycle management, including status checks and task cancellation.
- Detailed risk and confidence scoring allowing for nuanced agent decision-making logic.
- High-volume support for enterprise environments requiring more than 100 requests per day.
SKILL.md
Loading
Related Openclaw Skills
Ubuntu Linux Security Hardening Tool
A professional-grade configuration engine that generates customized security hardening scripts and policies for Ubuntu Linux systems.
Evernote and Yinxiang Integration
A lightweight REST-based integration for Openclaw Skills to manage Evernote and Yinxiang notes, notebooks, and tags.
ProjectPilot
A zero-dependency Python CLI tool for AI agents to manage projects, track tasks, and visualize progress through burndown charts and status reports.
Mapbox Geospatial Operations
A comprehensive decision framework for selecting the optimal geospatial tools within the Mapbox ecosystem based on performance, accuracy, and problem type.
Pretext + Three.js Viral Frontend Design
Build immersive web experiences where text functions as a physical, interactive material using high-performance rendering engines.
Gravitational Wave Event Analyzer
A powerful tool for fetching, processing, and visualizing real gravitational wave events from the GWOSC catalog.
Featured*
