Summary
Detect and exploit open redirect vulnerabilities across all parameter types, then escalate to OAuth token theft, SSRF, CSP bypass, and phishing.- Phase 1 discovers redirect parameters via passive recon (gau, waybackurls), active crawling (katana), HTTP header abuse, and OAuth/SSO-specific parameters
- Phase 2 verifies basic open redirect with a canary before reaching for bypasses
- Phase 3 covers 13 bypass technique categories: protocol substitution, allowlist suffix/prefix bypass, @ operator, URL encoding, double encoding, Unicode/IDN, path traversal, IPv6/IP confusion, parameter pollution, fragment tricks, case sensitivity, and CRLF injection
- Phase 4 targets OAuth
redirect_urithe highest-impact variant with bypass techniques and token capture setup - Phase 5 documents escalation chains: redirect → SSRF, redirect → CSP bypass, redirect → phishing, redirect → SAML/SSO token theft
- Phases 6–8 cover blind open redirect via OOB, automated testing with openredirex/ffuf/nuclei, and a complete report structure with CVSS vectors
SKILL.md file
Discover skill details
Discover skill details
Redirect Forge Open Redirect
What Does It Check?
In scope:- All redirect parameter names: redirect, return, next, goto, url, dest, forward, callback, and 20+ more
- HTTP headers that trigger redirects: Referer, X-Original-URL, X-Forwarded-Host, Host
- OAuth/SSO parameters: redirect_uri, callback_url, post_logout_redirect_uri
- 13 categories of bypass techniques for allowlists and WAF rules
- Escalation chains: OAuth token theft, SSRF, CSP bypass, CRLF injection, phishing
- Redirects to the same domain with no external navigation (not a vulnerability)
How It Works
Phase 1: Discover Redirect Parametershttps://evil.com.Phase 3: Bypass Techniquesredirect_uri with attacker server → victim completes OAuth → auth code delivered to attacker → account takeover.Chain with same-domain open redirect when the IdP whitelists https://target.com/*:Output
Known Limitations
- App that redirects to a fixed list of URLs is not vulnerable verify redirect destination is truly attacker-controlled
Locationheader present but middleware blocks external domains downstream → false positive; test with actual browser or curl -L

