How to use the WAF Bypass Techniques agent skill

What the skill makes the agent do

A web application firewall (WAF) sits in front of a website and analyses HTTP requests, blocking traffic that looks suspicious. This skill loads a checklist into the agent so that, during an authorized web application test, it works through known methods for getting test requests past that filter.

According to the skill's instructions, the agent follows the checklist in order unless you say otherwise, considers each technique against your target, tracks which items are complete, and suggests next steps based on findings. The checklist covers identifying which WAF is present (for example by cookies, headers, routes or JavaScript objects), encoding and header variations, case variation, comment injection, chunked encoding, IP considerations such as residential IPs, headless browser hardening, scraping APIs, and looking for origin servers that skip the WAF.

• Fingerprint the WAF using control pages, response headers, cookies and specific routes.
• Try encoding variations such as URL, HTML, Unicode and double encoding.
• Test header manipulation, chunked encoding and timing-based approaches.
• Consider residential or mobile IPs and headless browser libraries.
• Check whether an origin server can be reached without going through the WAF.

Install the skill (ZIP download or Claude Settings upload)

Download the skill ZIP from the agent-skills directory page. Skills are folders containing a SKILL.md file. For this skill, the only file listed is SKILL.md.

For Claude Code, unzip the download into your personal skills folder ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it available only in that project. Make sure the skill folder ends up containing SKILL.md.

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

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Claude apps: Settings → Capabilities → Skills

Trigger the skill with natural requests

You do not need a special command. The skill activates when your conversation involves terms such as WAF bypass, web application firewall bypass, URL encoding, double encoding, Unicode bypass, comment injection, HTTP header bypass, chunked encoding, IP rotation, payload obfuscation or WAF evasion.

State the authorization and target context in your request. The agent will apply the checklist to that context, so a clear scope helps it choose relevant techniques.

• "We have written authorization to test example-target.com. Its WAF blocks my test payloads. Which WAF is it and what checklist items apply?"
• "Our WAF rules should catch double-encoded payloads. Help me test whether they do in our staging environment."
• "Walk through the WAF fingerprinting steps for this authorized assessment and track what you have completed."

Worked example: a typical authorized test session

Suppose you are testing a web application under a signed scope agreement and a request is blocked. Start by asking the agent to identify the WAF from the block page, response headers and cookies. Once the vendor is suspected, ask it to go through the matching vendor-specific items and record which ones you have tried.

Keep the agent's findings in your own notes. The skill asks it to track completed items and suggest next steps, so you can review what was attempted before writing up results.

What you need before you start

You need a Claude-compatible agent that supports skills, and authorized targets with a scope agreement for testing. The skill's own documentation lists optional third-party tools that the checklist mentions, including undetected_chromedriver, Playwright stealth plugins, Cloudscraper, or a scraping API account such as ZenRows. You only need these if you choose to use those techniques.

The skill runs no scripts and uses no network access of its own. It is a set of instructions for the agent, so any traffic comes from the tools and testing you carry out.

• A Claude-compatible agent with skills support
• Written authorization and a scope agreement for the target
• Optional: undetected_chromedriver, Playwright stealth plugins, Cloudscraper, or a ZenRows account

Safety notes: scripts, network and legal scope

This skill is designed for offensive security testing and assumes you have authorization to test the target. Its methods are built to evade protective controls, so using them against systems you do not own or have permission to test may be illegal.

Do not use the origin-server and scraping techniques against production systems without coordination with the owner, because they can disrupt live services and affect third parties. The skill is not meant for simply fixing blocked legitimate requests; it focuses on evasion rather than allow-listing or tuning rules.

• Only test targets you have written permission to assess.
• Coordinate with the system owner before any production testing.
• Do not use this skill to fix false positives on your own legitimate traffic; tune the WAF rules instead.

Tips for getting useful results

Give the agent the WAF's block page text or response headers if you have them, because the checklist starts with identification. Ask it to follow the steps in order unless you have a reason to skip ahead, and to consider each technique against your specific target rather than trying everything.

If you are a security team validating your own WAF, ask the agent to frame results as detection gaps so you can write and verify new rules.

• Provide headers and block page details early.
• Ask for the checklist to be tracked so nothing is repeated.
• Frame findings as rule gaps when testing your own WAF.