All engineering notes
Web3, Growth, architecture record

Web3 Incentives & Quest Platform

Any incentive campaign with real value attached immediately attracts automated farming, and a naive implementation ends up paying thousands of wallets controlled by one person. The engineering here is mostly defensive: proof-of-personhood scoring, bot challenges and multi-wallet support are layered so that a participant has to be plausibly human before an allocation is recorded. The presentation layer is deliberately elaborate — a real-time 3D scene — because campaign products live or die on whether people want to be there.

49Source files, client application
3Anti-abuse layers, personhood, challenge, gating
2+Wallet ecosystems, EVM and non-EVM
Real-time 3DRendering, WebGL scene

Repository shape

Client application with 3D presentation, multi-chain wallet support and anti-abuse integrations

How this is presented

Documented as an architecture record. Campaign platforms are time-boxed by nature, so the anti-abuse design outlives any particular campaign.

Getting from a visitor to a verified allocation

Watch how many checks sit between arriving and earning. Each one exists because campaigns with real value attract automation.

Wallet layerExternal

Connects accounts across EVM and non-EVM ecosystems.

Client surfaceExternalServiceOn-chainData store
Completing a quest

Why the friction comes first.

Highlights

  • Layered anti-abuse: personhood scoring plus a bot challenge plus allocation gating, so no single bypass is sufficient.
  • Multi-ecosystem wallet support, letting a campaign reach users outside a single chain's community.
  • Real-time 3D presentation with post-processing, treating engagement as an engineering requirement rather than decoration.
  • Allocation records are written as verifiable facts rather than tallied in a spreadsheet.
  • Quest verification distinguishes on-chain actions, which are provable, from social actions, which are not.

Prove you are a person, then earn

The flow is deliberately front-loaded with friction: identity and personhood are established before any quest counts. That ordering is what stops a farming script from accumulating allocations across hundreds of wallets.

Presentation

A real-time 3D campaign surface with quest progress and reward feedback.

  • WebGL scene with post-processing effects.
  • Interaction feedback designed to make participation feel worth doing.

Identity and personhood

Wallet connection across ecosystems, plus evidence that a participant is human.

  • Personhood scoring integrated before allocations are recorded.
  • Bot challenge as a second, independent layer.

Quest verification

Confirms a quest was genuinely completed before crediting it.

  • On-chain actions are verified against chain state.
  • Social actions are treated as weaker evidence and weighted accordingly.

Allocation

Records what a participant has earned.

  • Written as records rather than accumulated in ad-hoc totals.
  • Gated behind the personhood layers.

Client structure

A single-page application where the anti-abuse integrations carry more weight than the feature surface — because a campaign with real value attached is attacked immediately. 16 units across 4 groups.

Presentation

5
pagescomponentsassets3D scenehooks

State & services

5
contextsserviceslibutilstypes

Chain

3
ABIsEVM walletsTON connect

Anti-abuse

3
personhood scoringbot challengefirebase

Stack by layer

Presentation

React Three Fiber, Post-processing effects, Tilt/parallax interactions, Confetti feedback

Identity

EVM wallets, Non-EVM wallet connect, Mnemonic accounts

Anti-abuse

Proof-of-personhood scoring, Bot challenge, Allocation gating

Backend services

Campaign state, Quest verification, Allocation records

Delivered work

Campaign platform

  • Real-time 3D campaign surface with post-processing effects.
  • Multi-ecosystem wallet connection.
  • Quest progress, completion feedback and reward presentation.

Anti-abuse

  • Proof-of-personhood scoring integrated ahead of allocation.
  • Independent bot challenge layer.
  • Allocation gating so rewards are recorded only after checks pass.
  • Differentiated verification for provable and unprovable quest types.

Key modules

Layered anti-abuse

Keeps allocations going to people.

  • Multiple independent layers rather than one check.
  • Ordered before earning, not after.

Quest verification

Distinguishes proof from claim.

  • On-chain actions verified directly; social actions weighted lower.

3D presentation

Engagement treated as a requirement.

  • Real-time WebGL rather than static marketing pages.

Presentation approach

How this project is presented

  • Anti-abuse design explained, since it outlives any single campaign.
  • No live campaign — these are time-boxed by nature.
  • Verified file counts.

Documentation and notes

Anti-abuse model

The layers, their ordering and what each one catches.

Quest types

How provable and unprovable completions are treated differently.
  • Campaign and organisation names are intentionally omitted.
  • Anti-abuse raises the cost of farming; it does not eliminate it, and the design assumes that.
  • Scale figures are counted from the repository.
All 21 engineering notes