How to use the offensive-xss XSS testing checklist skill

What the offensive-xss skill makes the agent do

The skill loads a methodology for XSS testing and asks the agent to apply it as an operational checklist. It covers stored, reflected, DOM-based and blind XSS, and it also describes polyglot payloads, CSP bypass, XSS filter bypass, event handler injection, DOM clobbering, mutation XSS and impact escalation such as session hijack, phishing and keylogging.

The skill is a methodology and reference. It does not attack anything by itself, and its source contains mostly explanations and example payloads with no scripts for the agent to run. A human still needs to judge results and test in a browser or a test environment.

• Find user input points such as URL parameters, form fields, HTTP headers, file uploads, JSON or XML inputs, WebSockets and API endpoints.
• Decide whether input is stored, reflected or handled in client-side code, and test each case accordingly.
• Confirm whether a payload actually runs in the browser, or whether a blind payload calls back to a server you control.
• If payloads are blocked, work through filter and Content Security Policy bypass ideas.
• Assess who an XSS could affect and how far it could be escalated.

How to install the offensive-xss skill

The skill is distributed as a ZIP file from the download link on this page. Its folder is named offensive-xss and its only listed file is SKILL.md. Unzip it into one of two locations, depending on whether you want the skill for all your projects or just one.

Option one is your personal skills folder at ~/.claude/skills/. Option two is a project-level folder called .claude/skills/ inside the project you are working on. Either way, the offensive-xss folder should sit inside the skills folder so that SKILL.md is found.

If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

• Download the offensive-xss ZIP from this page.
• Unzip it so the offensive-xss folder contains SKILL.md.
• Place the folder in ~/.claude/skills/ for personal use, or in .claude/skills/ inside a project.
• Or upload the skill in the Claude apps under Settings → Capabilities → Skills.

How to trigger the skill with natural requests

You do not need a special command. The skill is meant to activate when your conversation involves XSS topics. Its listed trigger phrases include XSS, cross-site scripting, stored XSS, reflected XSS, DOM XSS, blind XSS, CSP bypass, XSS filter bypass, polyglot, DOM clobbering, mutation XSS and event handler injection.

Write requests that say what you are testing and what you are allowed to test. Naming the target type, the input you are looking at and your goal helps the agent apply the checklist to your context.

• Walk me through an XSS checklist for the comment form on my own staging app.
• Which input points should I check for reflected XSS in this search page? I have written permission to test it.
• I think the profile page has stored XSS. What checks should I run to confirm impact?
• Our CSP is blocking inline scripts. Which CSP bypass ideas from the checklist apply here?
• Help me judge how serious a blind XSS in our admin log viewer would be.

Examples of how the checklist is used

A developer reviewing their own application before release can ask the agent to walk through the input points and identify where user data is stored or reflected. The checklist gives a clear order of work and a way to record which items are done.

A student can use the skill to compare how stored, reflected, DOM and blind XSS differ, using the diagrams and explanations in the methodology. A bug bounty tester can use it to keep a structured list of what has been tried on an in-scope target and what still needs checking.

What you need before using it

The listing states that the skill requires nothing beyond the agent itself. It does not run scripts and does not use the network. The checklist still expects you to have a browser or a test environment where you can observe whether a payload runs, and it expects human judgement to interpret the results.

The skill is not a one-click scanner. If you expect the agent to scan a site and return a finished report, this skill is the wrong tool.

• An AI agent that supports agent skills and the offensive-xss folder installed.
• A browser or test environment where you can observe behaviour, such as DevTools.
• A website or application you own, or written permission to test it.
• Security background sufficient to read the findings and decide what they mean.

Safety notes and tips

Only test websites you own or have written permission to test. The listing notes that using the skill against third-party systems may be illegal and violates most bug bounty rules. The skill runs no scripts and makes no network calls itself, so any requests made during testing come from you or the agent acting on your instructions.

The methodology's note on browser behaviour is useful: Chrome, Firefox and Safari may suppress alert, confirm and prompt dialogs in cross-origin iframes or background tabs. For more reliable detection, the source suggests side effects such as console.log, network beacons or DOM changes you can observe in DevTools.

• Keep a written record of which items on the checklist you have completed.
• Prefer side effects you can observe in DevTools over relying on alert dialogs.
• Ask the agent to suggest next steps based on what you found.
• Review the agent's output yourself before reporting or acting on it.