A security-first skill for AI agents to verify URL safety and intent alignment before navigation or data submission.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install urlcheck
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 urlcheck 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 URLCheck Security & Intent Verifier?
URLCheck is a sophisticated security companion designed for AI agents operating within the ecosystem of Openclaw Skills. It goes beyond simple blacklisting by analyzing target web resources for phishing, malware, and suspicious patterns in real-time. The skill acts as a critical gatekeeper, ensuring that any URL an agent encounters is thoroughly vetted before the agent takes further action like clicking, navigating, or downloading files.
Its standout feature is intent-aware verification, which evaluates whether a destination URL actually aligns with the user's stated goals. By integrating this into your set of Openclaw Skills, you empower agents to detect "mismatch" scenarios where a site might be technically safe but contextually wrong for the user's objective, such as a legitimate site that isn't the specific login portal the user intended to use.
URLCheck Security & Intent Verifier Use Cases
- Verifying links discovered during web research before an agent navigates to them.
- Validating login pages before an agent attempts to submit user credentials.
- Checking file download links for potential malware or security threats.
- Ensuring a redirect chain leads to a legitimate destination consistent with user intent.
- Assessing third-party payment or booking sites for alignment with the user's purchase goals.
How URLCheck Security & Intent Verifier Works
- The agent identifies a target URL and captures the user's specific intent (e.g., "log in to my account").
- The url_scanner_scan_with_intent tool is invoked to perform a deep analysis of the resource.
- The system evaluates the URL for both technical threats (risk score) and contextual alignment (intent alignment).
- A structured directive such as ALLOW, DENY, or RETRY_LATER is returned to the agent along with a detailed access reason.
- The agent follows the directive, either proceeding with navigation or warning the user based on the security findings provided by these Openclaw Skills.
URLCheck Security & Intent Verifier Setup
Install the required plugin using the CLI to enable these Openclaw Skills functionality:
openclaw plugins install @cybrlab/urlcheck-openclaw
openclaw gateway restart
Verify that the plugin and skill are active:
openclaw plugins list | grep -i urlcheck
openclaw skills list | grep -i urlcheck
Ensure the configuration plugins.entries.urlcheck-openclaw.enabled is set to true in your settings.
URLCheck Security & Intent Verifier Data Schema & Taxonomy
The skill returns structured payloads containing security metrics and access directives. Use the following fields to inform agent logic:
| Field | Description | Type |
|---|---|---|
| agent_access_directive | Primary action command (ALLOW, DENY, RETRY_LATER) | String |
| risk_score | Probability of threat (0.0 to 1.0; lower is safer) | Float |
| intent_alignment | Alignment signal (misaligned, no_mismatch_detected, etc.) | String |
| confidence | Certainty of the analysis (0.0 to 1.0) | Float |
| analysis_complete | Boolean indicating if the full scan finished | Boolean |
URLCheck Security & Intent Verifier Advanced Features
- Intent-aware verification to detect legitimate sites that do not match specific user goals.
- Non-blocking asynchronous workflows using task handles for handling long-running scans without freezing agent execution.
- Comprehensive task lifecycle management including status polling, result retrieval, and task cancellation.
- Detailed risk and confidence scoring for nuanced agent decision-making within the Openclaw Skills framework.
- Fallback mechanisms that prevent unsafe navigation if the verification service is temporarily unavailable.
SKILL.md
Loading
Related Openclaw Skills
Botcast
A transcript-first podcast platform enabling AI agents to engage in long-form, turn-based interviews as either hosts or guests.
OmniFocus Automation Skill
A powerful automation bridge for OmniFocus on macOS that enables programmatic task, project, and folder management.
GifHorse
A command-line tool that transcribes video libraries to enable dialogue-based searching and automated reaction GIF creation with timed subtitles.
Website Builder
An autonomous agent skill that researches market gaps, develops a web-based solution, and deploys it to Vercel without user intervention.
fluxA-x402-payment
A FluxA Agent Wallet skill that enables AI agents to autonomously request budgets, sign x402 payments, and call paid API endpoints.
ShipStation Order Monitor
A powerful monitoring tool for ShipStation that tracks new orders, flags processing delays, and automates fulfillment alerts for e-commerce businesses.
METADATA
Featured*
