Portaldot Hacker House 2026 | Katalyst: verifiable intent settlement on Portaldot

Portaldot Hacker House 2026 | Multi-Chain & Cross-Chain Infra

Project Introduction

Katalyst turns a plain-language intent into an on-chain route that nobody has to take on trust. GenLayer validators decide where funds should go, the Portaldot vault re-checks that decision on-chain before paying, and anyone can later verify the result.

The idea behind Katalyst is simple: when software moves your money, you should be able to see why, and check it yourself.

Katalyst is the decision and settlement layer that makes any aggregator’s choices verifiable.

A user states an intent and deposits funds. Validators agree on a route. The vault pays. A guardian can veto any proposal that doesn’t match the verdict, and a single command re-derives the whole story from chain events.

Intent in. Settlement out.

Problem Solved

Imagine you ask a system to put your funds where they earn the most, in A-Tier assets only. A model or a relayer decides where the money goes. You get a result, but nobody can show you why that route was chosen.

And what if the model was manipulated by a hidden instruction? Or one relayer sent a different route from the one that was agreed?

Most intent-settlement designs ask users to trust one of three things: hidden reasoning, a model that can be talked into mistakes, or a single relayer holding the verdict.

That’s the problem Katalyst addresses.

Katalyst splits the decision from the money. Validators decide, the vault re-validates on-chain, payouts are bound to a commitment the relayer can’t change, and a guardian can veto any mismatch.

It doesn’t ask users to trust the decision. It lets them check it.

Links

:globe_with_meridians: Live Application:

:clapper_board: Demo Video:

:computer: GitHub:

:bar_chart: Pitch Deck:

Technical Highlights

  1. Decision split from settlement

GenLayer validators read the intent, apply guardrails and agree on a route. The Portaldot vault then re-validates that route on-chain before it moves funds.

No single relayer or model holds the verdict.

  1. Commitment-bound payouts with a guardian veto

The relayer cannot change who is paid, how much, or the prompt. Settlement waits through a safety delay, and a guardian can catch and veto any proposal that differs from the validators’ verdict.

On testnet, a rogue proposal for intent #4 (a route the validators never chose) was proposed at block 136551, then vetoed with the depositor refunded in full at block 136564.

  1. Layered defence against prompt injection

A static screen, a constrained model and deterministic allocation sit in front of the vault, so a manipulated model cannot decide on its own.

In an earlier live run, a prompt-injection attempt was rejected by the solver and the deposit was refunded automatically.

  1. Verification anyone can run

Every settlement can be re-derived from on-chain events and GenLayer records. For intent #7, “katalyst verify 7” passes 12 of 12 independent checks, including commitment binding, exact split, route lock and safety delay.

Previously verified testnet evidence (intent #7):

  • Deposit: “0x1884a4aca29df239393eb7d5ea50438de5037c78d7aba3144a39a084644f81c5”, block 138211

  • Proposal: “0xff2239672d075f2bb30c1b355860293d364d3eef748c68bfcbdbdc405536aae1”, block 138218

  • Executed: “0xcaca1b520f7908283dd07dfb431f2a0d604c135677559f145a81ecbb597dd9c2”, block 138228

  • GenLayer consensus: “0xb7ca0f569ef71cbe72b401493c78b57b95a6929ecb72e3a3fbabea05521f7064” (Studionet, 3 of 5 validators agreed)

Further evidence includes the veto of intent #4 (“0x71498cf022ae9efff35fe2f4a0e7843c021f42a707e1795c77d6e6cb0d0c00fd”, block 136564) and a depositor cancellation with refund for intent #6 (block 136760).

  1. Built around Portaldot

The vault is deployed on Portaldot V3 Testnet (chain 420420777). KatalystVaultV2 is 5,704 bytes and PoolHub is 1,731 bytes. Because Portaldot charges per byte stored, the design keeps one storage slot per pending intent and forwards the storage-deposit refund to users. The Solidity contracts deploy for about 315 tPOT, against about 1,001 for the ink! variant.

  1. A simple, checkable fee model

Each settlement pays one fixed fee, 0.05 tPOT on testnet. The fee and the 50 tPOT deposit cap are fixed at deployment and cannot be changed, and “katalyst verify” checks the fee on every settlement.

  1. Working implementation and testing

Katalyst is publicly deployed, with its source code and documentation on GitHub. The release includes 11 contract scenario tests with randomised invariant fuzzing, and 31 solver and example unit tests.

Known limits: the pools on testnet are demo addresses, the contracts have not been audited, and no mainnet fee has been set. The roadmap targets each of these.

Team

Stunner and Kalila

Contact: @ernestam7 on Discord

Track: Multi-Chain & Cross-Chain Infra

Katalyst was built to make one part of on-chain automation less of a leap of faith: knowing why your funds moved, and being able to prove it.

Judges & Voting Fit

For Judges

Katalyst addresses a practical trust problem within the Portaldot ecosystem.

Rather than introducing another yield aggregator, it focuses on a defined layer: making the decision behind a cross-chain settlement verifiable, and keeping any single relayer or model from moving funds alone.

Its connected capabilities are decision-and-settlement separation, commitment-bound payouts with a guardian veto, and one-command verification.

The application is deployed, the source code is available, and its core workflows have been exercised on Portaldot V3 Testnet, including a rogue-relayer veto and a 12-of-12 verified settlement.

Why continued support matters: the next steps are a multisig and timelock, an external audit, a relayer quorum, real pool strategies and a mainnet launch with conservative limits. One solver can serve vaults on several chains, so each new chain adds settlement volume and $POT use on Portaldot.

For Community Voters

When software moves your money, the hard part isn’t seeing that it moved. It’s knowing why.

Katalyst is built around a simple belief: people should be able to check why a decision was made, not just trust that it was right.

It lets validators decide, makes the vault double-check, gives a guardian the power to stop a bad proposal, and lets anyone replay the result.

The project is early, and it is honest about what is still testnet. Its purpose is clear.

Katalyst — Verify, then trust.