How to use the Open Redirect Vulnerability Testing agent skill

What the skill makes the agent do

An agent skill is a set of instructions that Claude loads when a conversation matches the skill's topic. This one is a checklist for open redirect vulnerabilities, where a website sends visitors to an address chosen by the attacker. The skill's instructions tell the agent to apply the full methodology as its working checklist, follow the steps in order unless you say otherwise, track which items are complete, and suggest next steps based on findings.

The checklist covers where to look, such as redirect parameters like redirect, url, next and returnTo, plus login and OAuth flows. It also covers bypass ideas such as URL encoding, double slashes, domain spoofing and protocol handlers, and how a minor redirect can be chained into larger issues such as OAuth token theft or SSRF.

Install the skill from the ZIP download

Download the ZIP for the offensive-open-redirect skill from this page. Unzip it into one of two locations. For a personal setup, use ~/.claude/skills/. For a single project, use the .claude/skills/ folder inside that project.

Alternatively, you can upload the skill in the Claude apps under Settings → Capabilities → Skills.

• Personal: unzip into ~/.claude/skills/
• Project only: unzip into .claude/skills/ inside the project folder
• Claude apps: Settings → Capabilities → Skills, then upload

Trigger the skill with natural requests

You do not need a special command. The skill activates when your conversation involves its trigger phrases, which include open redirect, URL redirect, redirect bypass, URL encoding bypass, CRLF, protocol handler, redirect chain, OAuth redirect, SSRF chain and open redirection. Describe what you are doing in plain words that include one of these terms, and state that you are authorised to test the target.

Once active, the agent works through the checklist rather than answering in general terms, and it will suggest next steps after each finding.

• Review the redirect parameters on this login page for open redirect issues. I have written permission to test this app.
• My OAuth authorize endpoint may accept any redirect_uri. Walk me through the redirect_uri tests from the checklist.
• I found a redirect that blocks https://evil.example. What bypass techniques from the checklist should I try next?
• Can this open redirect be chained into an OAuth redirect chain or an SSRF chain? Suggest next steps.

Examples of what the checklist covers

The methodology lists common redirect parameter names and asks the agent to test them with absolute URLs, protocol-relative URLs and path traversal. For OAuth, it asks the agent to check whether redirect_uri validation is improper. For bypasses, it lists domain spoofing, encoding, protocol confusion and path-based variants.

These are the kinds of test strings the checklist contains. Use them only against systems you own or are permitted to test.

• Absolute URL test against a redirect parameter
• Protocol-relative URL test using double slashes
• Domain spoofing patterns such as a host that begins with the target's name
• Encoded scheme variants for a blocked redirect

What you need before you start

The skill lists no required tools, does not run scripts and does not use the network itself. It consists of a single instruction file, SKILL.md. Any network requests during testing come from the tools and environment you use, so the testing itself depends on your own setup and permissions.

Written permission to test the target is the key requirement. The skill is intended for web apps you own or are authorised to test, and it is not intended for non-security tasks or general redirect building.

• Requires: nothing listed in the skill
• Runs scripts: no
• Uses network: no
• Permission to test the target in writing

Safety notes and limits

The content is attack-oriented. Probing systems you do not own or have written permission to test can be illegal and harmful, so keep every session scoped to authorised targets. The skill does not give general help with building redirects.

Because the skill itself runs no scripts and makes no network calls, you control what gets sent to a target. Review each suggested request before you run it.

• Only test authorised targets
• Review any request the agent suggests before sending it
• Do not use the skill for non-security redirect work

Tips for better results

Give the agent context: the page or endpoint, the parameter names you see, and any blocking behaviour. The checklist asks the agent to consider applicability to the current target, so specific details produce more useful next steps. Ask it to track which items you have completed so you can see coverage at a glance.

For OAuth targets, mention the flow you are testing. The source notes that many OAuth and SSO stacks require an exact redirect_uri match, so the agent will look for partial or path-only allowlists and case or encoding mismatches.

• Name the endpoint and parameter you are testing
• Share exactly what the server returned when blocked
• Ask for the checklist order, or say if you want to skip ahead