| name | auth-and-session |
| description | Authentication and session-management technique list (brute-force rate-limit bypass, JWT, OAuth/OIDC, SAML/SSO, session fixation, password reset, MFA bypass) plus 2025-26 modern identity attacks (passkey/WebAuthn, OAuth device-code phishing, AiTM relay, ghost logins). Converted from master-pentest-prompt.md Phases 7/7.5. Use on any login, session, SSO, or identity-related flow. |
Auth & session -- everything
When to use
Any login form, session-handling code path, SSO/OAuth integration, or
password-reset flow.
Core auth & session techniques
- Brute-force: rate-limiting bypass by rotating
X-Forwarded-For,
X-Real-IP, X-Client-IP, True-Client-IP, spoofed User-Agent, IPv6,
case variation.
- JWT: algorithm confusion (
none, HS/RS), kid injection,
jku/x5u external reference, weak-secret brute force, claim tamper
(role/admin/exp), JWKS confusion.
- OAuth/OIDC:
redirect_uri open redirect, state parameter
leakage/missing, code swap, token mix-up (SSRF), consent-screen
phishing.
- SAML/SSO: XML signature wrapping, missing signature validation,
forged XML, replay, login-as via injected attribute.
- Session: fixation, predictable token, missing flags
(
HttpOnly/Secure/SameSite), concurrent sessions, session not
invalidated on logout/password change.
- Password reset: token leakage via Referer, predictable token, Host
header alternate injection (
X-Forwarded-Host, absolute URL in body).
- MFA bypass: OTP replay, race conditions, backup-code enumeration,
MFA missing on the API path, cache-based bypass, "remember me" device
cookie forgery. See the expanded MFA bypass list below for the full
arsenal beyond these basics.
- Email/SMS: notification bombing, verification bypass, magic-link
token brute force.
- Account takeover chains: username -> cookie manipulation
(
Cookie: user=admin).
- Login CSRF, OAuth CSRF, password-change CSRF.
- Cache poisoning -> session cookie disclosure.
Modern identity & browser-era attacks (2025-26)
- Passkey/WebAuthn: FIDO2 URI intent hijacking (CVE-2024-9956 class),
passkey downgrade to password, WebAuthn CSRF (add a passkey with the
victim's email), duplicate passkey login confusion, discovery-order
preference abuse.
- OAuth device-code flow: device-code phishing (the Salesforce-breach
class), leak of
user_code via open redirects, polling-timeout window
abuse.
- Consent phishing / malicious OAuth app approval; state-less OAuth.
- AiTM / reverse-proxy phishing exposure: does the app's MFA actually
stop an adversary-in-the-middle relay session? Reverse-proxy phishing
kits (Evilginx3, Modlishka, Muraena) clone the login portal and proxy
every request live, harvesting the session cookie right after the
victim clears MFA -- since most apps only check MFA at authentication,
not per-session, the stolen cookie alone is enough afterward. Test by
replaying a stolen session cookie after MFA completes.
- ClickFix / FileFix / ConsentFix: HTML injection or open redirect
that could be weaponized for browser-native social engineering.
- Malicious browser-extension vector: exposed
/manifest.json or
update endpoints; extension permission overreach in the supply chain.
- Ghost logins / SSO coverage gaps: flows that bypass SSO entirely
(legacy endpoints, partner portals, API-key fallback) -- enumerate
unsecured auth paths alongside the SSO path.
- Session device binding: can a session cookie be replayed on a
different device/browser/ASN without a re-prompt?
Expanded MFA bypass list
Beyond the basics above, these are the specific classes worth checking on
any MFA-protected flow -- distilled from a 2026 red-team/bug-bounty survey
of what's actually working against "unbreakable" MFA in practice:
- MFA fatigue / push bombing: repeatedly trigger login attempts with
already-obtained credentials so the victim's phone gets flooded with
push approval prompts, betting they'll eventually tap "Approve" just to
make it stop. Defense signal to check for: number-matching push
(approve requires typing a code shown on the login screen, not just
tapping yes/no) or FIDO2 hardware tokens, which this attack can't touch.
- Legacy authentication protocol bypass: IMAP/POP3/SMTP/SMB and
similar legacy protocols often don't enforce MFA at all even when the
main web login does --
nmap -p 143,993,110,995,445,<target> to check
what's exposed; a valid password alone can grant mailbox/file access
through the legacy path even with MFA fully configured on the web app.
- MFA reset via helpdesk social engineering: call/chat into support
claiming a lost device, using pretexting (urgency, impersonating a VIP)
to get MFA reset or removed. Worth probing for during a social-
engineering-in-scope engagement, and worth flagging as a policy gap
even when out of scope to test live (does the org require strong
identity verification and out-of-band confirmation for MFA resets?).
- Self-enrollment / registration loopholes: many apps let a user
register a new MFA factor (phone, authenticator) with minimal
re-verification, especially right after a "lost device" flow --
test whether an already-compromised session/account can add an
attacker-controlled MFA factor and use it for ongoing access.
- Pass-the-cookie / pass-the-token: extract a valid session
cookie/token (via XSS, endpoint malware, or MiTM) and replay it in a
fresh browser profile or via an automated client -- MFA doesn't
re-prompt as long as the cookie itself is valid, so this bypasses MFA
entirely rather than attacking it directly. Check whether the target
binds sessions to a device fingerprint or IP, since many still don't.
- SIM swapping / phone number hijacking: relevant context for
SMS-based MFA specifically -- if SMS OTP is the only second factor
offered, that's worth flagging as a weaker option than an
authenticator app or FIDO2 regardless of whether SIM swapping itself is
testable in a given engagement (it targets the telco, not the target
application).