Skip to content
Fuga Labs
Scope an engagement

The perimeter. Five walls, one substrate, zero custody.

Any one of these can be built by a competent team. The compounding cannot be bought a piece at a time. One signing and attestation model, one provenance discipline, one response ladder, one custody-free posture, one Circuit Breaker contract that belongs to you.

Scope an engagement Read the record
Wall 01

Live Detection

See

The sensor layer the rest of the perimeter reads, and the only wall that can stop a loss rather than report one. It watches what a transaction is attempting rather than what it resembles, works out what would happen if it landed, and hands the answer either to your on-call or to a Circuit Breaker contract you own.

  • Pre-inclusion visibility
  • Outcome simulation
  • Seven-rung response ladder
  • The Circuit Breaker, owned by you

Is an attack executing on-chain right now, and can it be stopped before it lands?

Live Detection
Wall 02

Ops Monitor

Sense

Contract audits are necessary, expensive and mostly already done, and they do not touch any of this. Ops Monitor watches the organisation around the code: who can sign, what a proposal really executes, where your domain resolves, and whether the endpoint your infrastructure trusts is telling it the truth.

  • Signer and threshold changes
  • Governance read from calldata
  • Frontend and DNS
  • RPC integrity

Is this protocol being operationally compromised right now?

Ops Monitor
Wall 03

Exposure Map

Map

You think you hold one asset. You actually hold a wrapper, backed by a staked position, sitting in a lending market, borrowed against. Exposure Map follows that chain of claims to the bottom and tells you what an incident anywhere along it is worth to you.

  • Look-through exposure
  • Real exit price
  • Contagion order
  • Proof of backing

If asset X is compromised, who is downstream and for how much?

Exposure Map
Wall 04

Asset Ratings

Certify

Before you accept an asset as collateral, somebody should ask how many independent parties must agree before new units of it can appear, and whether anything is verifiably behind it. Asset Ratings asks, publishes the answer as a grade, and gives you one contract line that refuses anything below the grade you set.

  • Public methodology
  • A floor, not an average
  • Watched for drift
  • Enforced at onboarding

Machine-readable minimum standards for what you build on.

Asset Ratings
Wall 05

AI Agent Guardrails

Bound

Software is starting to move real money on its own: rebalancing, paying, executing strategies, calling contracts on somebody's behalf. Today the thing standing between a hostile input and a treasury is usually a prompt. This replaces the prompt with limits a contract enforces before the transaction executes.

  • Per-operation caps
  • Rolling velocity limits
  • Allowlisted targets
  • A kill switch you hold

Which AI agent is this, who operates it, and what is it allowed to do?

AI Agent Guardrails

Why the five compound.

Read these as arrows rather than a list. Every wall makes the next stronger and the ones underneath it cheaper, which is the part a competitor cannot answer by shipping a feature.

Live Detection everything

The sensor layer, and the Circuit Breaker the other walls borrow their enforcement contracts from. Every wall consumes its findings or its evidence, and the wallet lane sees intent at scale, which is the richest signal in the stack and the cheapest to reach.

Exposure Map Live Detection

Blast radius. Without it a response ladder cannot exceed notify, because you cannot act proportionally on a number you do not have.

Asset Ratings Exposure Map

Structural weights. Contagion through a fragile asset is not the same as contagion through a robust one, and the graph has to know the difference to be worth reading.

Asset Ratings Live Detection

A reason to run tighter thresholds exactly where the structure is weaker.

Ops Monitor Live Detection

The half of the attack surface a chain view cannot see, on the same alert channel and the same rotation.

Live Detection AI Agent Guardrails

The bounded-authority contracts, already built and already owned by the customer.

Each wall has a different buyer.

So you land in one and expand into the others without a second procurement fight: the integration, the contracts and the alert channel are already there.

WallWho signsWhat they are buying
Live Detection, protocol lane Security lead, risk committee Knowing during, instead of after
Live Detection, wallet lane Head of product Users who do not get drained, which is a retention feature rather than a cost centre
Live Detection, the Circuit Breaker CISO, infrastructure Refusal at their own boundary, under their own control
Ops Monitor The same security lead The attacks that do not look like attacks
Exposure Map Risk and treasury A collateral parameter they can defend in a meeting
Asset Ratings Risk committee A gate, in their own code, with an argument behind it
AI Agent Guardrails Platform and engineering Being allowed near real money at all

Start with the wall that answers your worst week.

Every engagement is scoped to your protocol by the team. Pricing follows the scope and is settled before anything is signed.

Scope an engagement Find your sector