A dispute resolution system for Realms DAOs on Solana
Adds a safety layer to DAO treasury payouts. Instead of payouts executing immediately after approval, they enter a dispute window where community members can challenge suspicious transactions before funds leave the treasury.
The problem: DAOs need a way to catch malicious or mistaken treasury proposals before money is lost.
The solution: A challenge period with economic incentives (bonds) and neutral arbitration (AI or human resolvers).
- Safe Treasury Program — on-chain Solana program managing payout queues, challenges, and rulings
- Realms UI Integration — treasury management screens and proposal instructions built into the Realms interface
- AI + Human Arbitration — automated and manual dispute resolution workflows
A DAO votes to send funds (or even NFTs). Instead of immediate execution, the payout enters a queue with a dispute window (e.g., 24 hours).
That is made possible because the DAO's treasury is held in a safe treasury vault rather than the ordinary treasury vault of Realms DAOs. The safe treasury receives instructions from DAO governance, but has a time period open for requesting arbitration if the transaction does not respect the DAO's Constitution. DAOs wanting to adopt this safety mechanism must opt-in / migrate their treasury.
Anyone holding the required token can challenge the payout by posting a bond. This freezes the payout and triggers arbitration.
A designated resolver (which is a DAO of AI agents or human arbitrators, onboarding arbitral bodies to web3) reviews the challenge and issues a ruling: Allow or Deny.
- Allow → Payout executes, challenger loses bond
- Deny → Payout cancelled, challenger gets bond back
Screenshots captured by automated browser agent running against local test network. Better quality and demo to be updated.
The treasury screen shows:
- Safe Policy status (dispute window, bond requirements)
- Payout queue with challenge status
- Action buttons: Queue Payout, Exit Custody
DAOs can create proposals using new instruction types:
- MigrateToSafe — set up dispute policy (bond amount, dispute window, resolver address)
- QueuePayout — add a payout to the queue (SOL or SPL tokens)
- ChallengePayout — challenge a queued payout
- ExitFromCustody — withdraw funds from safe custody
✅ On-chain program — Safe Treasury program builds and deploys
✅ Localnet demo — Full bootstrap script creates test DAOs and payout fixtures
✅ UI integration — Instructions render in Realms proposal interface
✅ Treasury screens — Dispute Safe custody panel displays in treasury view
⏳ End-to-end flow — Full challenge → ruling → release cycle in browser (currently script-based) ⏳ Production readiness
| Repository | Core Feature Categories | Test Types | Status |
|---|---|---|---|
| arbitration-layer (Safe Treasury) | • Policy initialization & updates • Treasury registration & management • Native/SPL vault operations • Payout queue/release lifecycle • Challenge bond mechanisms • Ruling record/appeal/finalize • Governance proposal integration |
• Unit tests (state enums) • On-chain program tests • Formal verification (Certora) • Localnet script validation • Integration flow tests |
✅ Working |
| ai-arbitration-dao | • CLI command orchestration • Agent runtime & health monitoring • Proposal creation/voting/execution • Resolver binding & authorization • Ruling payload validation • Governance adapter integration • Observability & logging • Worker environment reconciliation |
• Unit tests (18 test modules) • Integration tests • CLI contract validation • Runtime environment tests • Secret redaction tests • JSON output validation |
✅ Working |
| human-arbitration-dao | • Realms DAO bootstrapping • Role-based access control • CLI interface for operations • Web voting interface • Case management workflows • Treasury operations |
• CLI integration tests • Web interface tests • Role model validation • Bootstrap deployment tests |
✅ Working |
| governance-ui | • DisputeSafe instruction registration • Treasury account store management • Transaction state handling • Router history tracking • Delegator selection • Vote stake registry • Proposal creation UI |
• UI component tests • Store state tests • Instruction registration tests • SSR page rendering (all routes) |
✅ Working |
- Total Test Files: 60+ across all repositories
- Core Flows Validated: Treasury operations, Arbitration workflows, Governance integration, DAO bootstrapping
arbitration-layer/: on-chain program tests, localnet script validation, and formal verification WIP artifacts undercertora/.governance-ui/: UI and integration-oriented test coverage for Realms arbitration surface changes.ai-arbitration-dao/: unit/integration/e2e scaffolding and command-level validation paths for agent runtime flows.human-arbitration-dao/: CLI modules with test coverage and executable scripts for arbitration operator workflows.
- Authority-First Orgs (strong onchain authority) – DAO Constitution and policy enforcement plus AI/Human arbitration DAOs demonstrate onchain authority and escalation paths for disputes.
- Governance Builders (tooling + governance systems) – Safe Treasury program to govern executable payout challenge flows for Realms DAOs.
- Realms Extensions (integrations/plugins) – governance-ui modifications slot arbitration features directly into the existing Realms surface (in v1, but happy to integrate in v2).
arbitration-layer: Core Realms Arbitration Layer workstream (program + deployment tooling WIP).governance-ui: Realms UI v1 fork/extension surface for arbitration support.ai-arbitration-dao: DAO scripts/runtime scaffolding for AI arbitrators.human-arbitration-dao: Human arbitrator DAO tooling (CLI).
| Program | ID |
|---|---|
safe-treasury |
9yMpZraAc4pFvg4DXTT3rhvUvdh2xGQUdiNLQ1bwEhCD |
spl-governance (mainnet binary) |
GovER5Lthms3bLBqWub97yVrMmEogzX7xNjdXpPPCVZw |
# Terminal 1 — start validator (leave running)
bash arbitration-layer/scripts/localnet/start-local-validator.sh
# Terminal 2 — bootstrap + seed + demo
bash demo-hackathon.shInstead of running the above, you can check a sample output in demo-hackathon-output.txt.
This is an early prototype and should not be placed into production without thorough vetting/testing/auditing.

