← marwandiallo.comlabs

Identity Lab

A hands-on playground for the auth model that's replacing passwords: passkeys, phishing-resistant MFA, and the workload-identity primitives that extend the same guarantees to AI agents.

0FIDO2 passkey logins have been successfully phished in the wild at scale. The origin binding makes it impossible.
~80%of breaches still start with a stolen or phished credential (Verizon DBIR).
1 originis all a passkey will ever sign for. That's the whole trick.

Try it

Passwordless sign-in

Register a passkey on your device, then sign back in with no password and no OTP. Walks through every byte of the WebAuthn ceremony — including the origin binding that makes it phishing- resistant.

Open demo →

JWT inspector

Paste a real token, see what's inside, and watch the analyzer flag alg=none, alg-confusion, missing exp, and PII leakage in real time.

Open inspector →

Phishing-resistant MFA, explained

Why FIDO2 / WebAuthn is the only mainstream MFA factor that survives a real-time AitM phishing kit. With diagrams of what actually happens on the wire.

Read →

Agent identity

What a passkey is to a human, a workload credential is to an AI agent. Covers OIDC, SPIFFE/SPIRE, attestation, and the new problem: how do you authenticate an agent that acts on a user's behalf?

Read →

Why this exists

I spend my days auditing identity systems at Microsoft. The same bugs keep showing up: tokens accepted with jwt.verify(token, secret) without an algorithm pinned, MFA prompts that any reverse-proxy phishing kit can replay, and now — agents handed long-lived API keys because nobody designed an identity model for them.

This lab lets you touch the alternative: phishing-resistant authentication for humans, the same primitives extended to workloads, and the failure modes that cause both to be deployed insecurely.

Modeled on real guidance