background gif
eni6ma

Attacks · Overview

Attack conversion

Most attacks succeed by turning something you can see or steal into something you can use. One-shot proofs break that conversion: surface observations need not become secret knowledge, and spent nonces cannot return as authority.

~7 min readContinue: Attack map

Opening

What the attacker wants

The attacker’s job is almost never “break math for sport.” It is conversion: take a transcript, a stolen token, a vault dump, a relayed ceremony fragment, or a mutated request and turn that capture into usable request authority. Classical estates make that step cheap. If possession is authority, then seeing or stealing something is already most of the win — the rest is packaging the secret as a caller. Every family in this catalog is really an argument about whether that conversion still pays. Buyers should read every family page as a specialized conversion story: different trays, same question — does capture become lasting authority?

How it works today

In the ambient world the conversion works by design. Bearer tokens authorize until expiry. API keys work from any host that holds them. Passwords and OTPs are strings you can type again. Session cookies follow the browser. Vaults, agents, and CI runners become blast-radius amplifiers because the same object that authenticates also travels. Capture and reuse are the same game: the notebook that fills with surface bits is also the tray that mints lasting callers. That is why vault dumps, phished OTPs, and replayed bearer tokens feel like the same incident class in postmortems: the object that authenticated also traveled.

Why it fails against one-shot proofs

ENI6MA aims to break the conversion step itself. Under empty-channel design, notebooks can fill with surface bits while effective mutual information with the secret stays empty under the named model. Request-bound envelopes and burn-before-validate then deny reuse even when an observer records every visible step. Seeing or stealing something is not the same as owning lasting authority. The plates below show the ambient conversion path and the denial stack — empty effective bits, spent nonces, and binding — that make family-level attacks fail to cash out. Absolute claims stay model-scoped to the reference architecture; the catalog teaches the conversion break without promising that observation is impossible.

See or steal → use

How to read: left capture trays feed a conversion path; right ambient win shows authority gained. Takeaway: classical attacks pay at the conversion step.

That left-to-right story is how classical breaches pay: capture trays feed a conversion gear, and the gear prints authority. If the gear turns, every downstream family is just a different tray.

Attack conversion fails

How to read: left conversion myths crossed out; right denial stack (empty MI, burn, binding). Takeaway: surface fills notebooks; authority does not transfer.

When conversion cracks, the rest of the catalog becomes mechanism detail — which check denies which tray — not a promise that nothing bad can ever be observed. Surface notebooks may still fill; the question is whether those surface bits become effective bits against the secret.

Scoped claims

Authority expires at the end of each request.ShippingThe envelope binds method, endpoint_id, request hash, policy hash, tau, and nonce (one message, one use).Holds under the reference architecture

Every envelope is bound to a specific method, endpoint, request body, policy, freshness window, and nonce.Shippingsha256(method || endpoint_id || request_body_hash || policy_hash || tau || nonce_uuid)Holds under the reference architecture

Where it shows up

  • APIs & workloads

    Stolen API keys become lasting callers — conversion is the whole breach story.

  • Agents & MCP

    Tool secrets that survive clone/copy convert into blast radius.

  • Human authentication

    Harvested passwords and OTPs are reusable credentials by design.

Watch it fail

Go deeper

  • Why Naive Attack Intuition Is Wrong

    Written for challengers who worry that watching multiple successful logins reveals the password and private map, or that random guessing eventually hits both, this essay explains why those paths do not work for the ENI6MA rotating-ring method (not mathematically, statistically, computationally, or physically).

  • Toward Perfect Secrecy by Design

    This paper presents an end-to-end security argument for an ENI6MA-style interactive proof in which the attacker gains nothing in Shannon’s sense of perfect secrecy. The mechanism combines a one-time, entropy-dependent projection of the alphabet into six perceptual leaves; ring-rotation diffusion that spreads symbol appearances across a large configuration space; confusion from a private bijective map to six synonyms; and a second-layer enumeration of the 6! leaf permutations.

  • Information-Theoretic Boundaries of the Rosario Primitive: Observable-Channel Isolation

    This paper formalizes information-theoretic boundaries of the Rosario Primitive with emphasis on the observable channel and adversarial inference limits. The central claim is that, in the ideal regime, the observable channel carries zero mutual information about the hidden state beyond a nonce-scoped event-validity residue; in the non-ideal regime, leakage remains small, measurable, and non-accumulative under proper constraints.

  • Attack Vectors Are Obsolete

    A practitioner-facing essay arguing that classical cryptanalytic and credential-harvest attack vectors lose their conversion path against the Rosario–Wang Cipher when there is no reusable private key on the wire. Frequency analysis, known/chosen plaintext, ciphertext-only, differential and linear cryptanalysis, and side-channel harvest are addressed as families that fail when the channel exposes membership-only residues rather than transferable secrets.

Formal note

Conversion denial is the visitor catalog frame for empty effective mutual information plus one-shot request authority: an observer may record surface bits while I_effective with the secret stays empty under the named model, and burned envelopes cannot return as lasting callers. Absolute claims remain model-scoped to the reference architecture. Go deeper via naive-attack-intuition and observable-channel papers for surface-versus-effective and residue framing.