Skip to content

adidshaft/red-dead-redemption-agentic-era

BranchesTags

Repository files navigation

Red Dead Redemption: Agentic Era

Public repo: github.com/adidshaft/red-dead-redemption-agentic-era

Player guide: docs/gamers-guide.md

Lightweight western arena game built for X Layer and OnchainOS. Players create named agents, fund skill upgrades on X Layer, switch between manual and autonomous control, and settle paid match outcomes onchain. The current product focus is an agentic gameplay loop: autonomous riders fight, rotate for supplies, play the shrinking ring, stage budget-aware follow-up actions, propose their next upgrades, and route players toward x402-powered premium autonomy. The live build now rotates across three real arenas, and map props are physically solid instead of decorative.

What Is In The Repo

  • apps/web: Next.js 15 client with wallet auth, roster management, X Layer skill purchases, and a Phaser 3 arena.
  • apps/server: Fastify + Socket.IO backend with wallet-signature auth, Postgres persistence, autonomous decisioning, queueing, and X Layer receipt reconciliation.
  • packages/shared: shared ABI, game rules, chain config, and API schemas.
  • packages/contracts: Hardhat contract package for ArenaEconomy and its tests.

Architecture Diagram

flowchart LR
  Player["Player"] --> Web["apps/web\nNext.js + Wagmi + Phaser"]

  Web <--> Server["apps/server\nFastify + Socket.IO"]
  Web <--> Shared["packages/shared\nABI + schemas + game rules"]
  Server <--> Shared
  Contracts["packages/contracts\nArenaEconomy"] --> Shared

  Web -->|Wallet signature auth| Server
  Web -->|Register agent / buy skill / enter paid match| XLayerTestnet["X Layer Testnet\nCore game economy"]
  XLayerTestnet -->|Receipts / explorer links| Web
  XLayerTestnet -->|Event verification / settlement reconciliation| Server

  Server <--> Postgres["Postgres\nAgents + matches + receipts + budget policy"]
  Server -->|Queue + match sim + AI logic + live snapshots| Web

  Web -->|x402 payment client| X402["x402 on X Layer Mainnet\nPremium autonomy pass"]
  X402 -->|402 challenge / payment settlement| Server
  Server -->|Premium autonomy status + receipts| Web
Loading

The web app handles wallet connection, signed auth, contract writes, and live arena rendering. The server owns queueing, match simulation, AI decisions, campaign state, budget-policy follow-ups, and transaction reconciliation. Core gameplay economy actions run on X Layer testnet, while the premium autonomy lane is unlocked through x402 on X Layer mainnet.

Live Network Snapshot

ArenaEconomy on X Layer Testnet

  • Contract address: 0x31a44d5dcA53A0BFB13C79d8dF5ED3148f08DB97
  • Deployment tx: 0xf6573f85ca2dfdc1e4cfee1a027782a1c620d918e3ce984280c12dacb268386a
  • Deployer: 0x1620e0aEd0215Add62EF51A109De35455D35B6A1
  • App treasury: 0x1620e0aEd0215Add62EF51A109De35455D35B6A1
  • Network: xlayerTestnet
  • Chain ID: 1952
  • RPC: https://testrpc.xlayer.tech/terigon
  • Explorer: OKX X Layer Testnet Explorer
  • Deployment block: 25006326
  • Deployed at: 2026-03-14T05:12:43.260Z
  • Artifact: packages/contracts/deployments/xlayerTestnet.json

Premium x402 Lane on X Layer Mainnet

  • Chain ID: 196
  • RPC: https://rpc.xlayer.tech
  • Explorer: OKX X Layer Mainnet Explorer
  • Default asset from .env.example: USD Coin
  • Default asset address: 0x74b7f16337b8972027f6196a17a631ac6de26d22
  • Default quote amount: 1000000 base units

Captured Proof Transactions

The longer-form proof context and reproduction notes live in docs/proof.md.

Core Features

  • Agent creation with BaseName-<6 char ULID suffix> naming.
  • Five core skills: Quickdraw, Grit, Trailcraft, Tactics, Fortune.
  • The skill UI now shows the live stat value, on-hover gameplay tooltip, and the exact combat effect each upgrade is currently contributing.
  • Starter skill distribution of 20/100 in each stat plus 10 random bonus points.
  • Manual or autonomous combat in a 4-agent free-for-all arena.
  • The web client now keeps the live arena and next-step guidance upfront, while deeper autonomy, onchain, history, and spectating tools sit behind a calmer tabbed operations console.
  • Queueing now has a live safety net: the client polls queue status, recovers missed live matches, and syncs active match snapshots so local play is less likely to feel stuck if a websocket update is missed.
  • The queue lane now reads as a short run-up flow with bot-fill progress, rider-mode guidance, and a clearer “opening bell” state instead of a generic wait message.
  • Queue fill now resolves faster and exposes live slot/ETA context, so a player can actually tell how many riders are locked and when house bots are due.
  • The arena HUD now surfaces one live battle directive, danger chips, ring pressure warnings, and nearest-threat cues so a player can understand the match state without reading a wall of text.
  • Field Intel is now centered on one “what matters now” call, a compact minimap, and a short plain-language legend so the panel stays readable instead of turning into a text wall.
  • The live arena now also carries a Town Pulse panel, landmark plaques, a stronger selected-rider beacon, and clearer intent cards so the player can tell what their rider, the ring, and the current side-prize are doing at a glance.
  • The arena now supports a rider-follow camera as well as a wide town camera, making the selected cyan YOU rider much easier to track during live fights.
  • The selected rider marker now stays consistently cyan on both the arena and the minimap, reducing “where am I?” confusion once the fight opens.
  • The Dust Circuit arena has been rebuilt as a live frontier scene with saloon, hotel, wash, stable, wagon, water tower, richer rider silhouettes, and stronger in-canvas combat cues instead of a flat prototype board.
  • The frontier landmarks now provide real cover, changing hit chance, damage mitigation, bot pathing, and the live combat guidance shown to the player.
  • Dust Circuit props are now physically solid, so riders cannot phase through saloon fronts, wagons, fences, crates, towers, or corral structures.
  • A second arena, Deadrock Gulch, now rotates into live matches with its own landmark graph, caravan lanes, and collision layout.
  • A third arena, Ironwood Crossing, now rotates in with depot, mill, freight-line, and stock-pen cover so the frontier feels denser and less repetitive.
  • Player and spectator guidance now stays short: the wallet briefing, rider setup, Field Intel, and Live Frontier board all use tighter copy and fewer repeated panels.
  • The post-match dossier now explains the rider’s finish, combat output, treasury outcome, and next recommended action instead of only showing raw standings.
  • The post-match dossier now also shows what changed for the rider after the run: next recommended skill, next queue type, campaign state, and a short recent-run tape.
  • The result overlay now also shows a career pulse, so every finished run feeds back into tier, streak, payout pressure, and a clear reason to queue again.
  • Live signal-drop objectives now appear during rounds to force convergence and reward agents who control tempo inside the ring.
  • Live bounty marks now spotlight the hottest rider in the field, turning mid-match momentum into an obvious kill target with bonus score and stronger on-canvas guidance.
  • Stagecoach runs now cut across town as moving score/ammo events, giving the field a visible mid-round prize that feels more like a living frontier than a static arena.
  • Autonomous combat behavior includes targeting, ring rotation, pickup routing, reload timing, and fallback survival logic.
  • An Autonomy Director surfaces each agent's doctrine, next skill target, economy loop, and x402 upgrade path.
  • The Autonomy Director now gives every rider a compact mission, a three-step agenda, and a campaign hook so the AI loop reads like an operator plan instead of raw planner text.
  • The autonomy lane is now framed as plain-language Autopilot: it focuses on when the rider takes over, what it controls in the fight, what still requires the owner, and what premium x402 adds without planner jargon first.
  • The rider panel now makes the Manual vs Autopilot split explicit: players still approve skills and queue entry, while Autopilot only takes over once the live match starts.
  • Riders now carry budget guardrails for autonomous follow-up actions, including an upgrade budget, a single-buy cap, paid-entry reserve, and a queue posture that keeps the economy loop understandable instead of opaque.
  • Budget-aware Autopilot can now tee up the next skill buy after a finished autonomous run and optionally roll the rider straight into another practice match when the policy allows it.
  • Live autoplay calls are now simplified and surfaced both in the arena and in the Autopilot console, so the agent’s current decision reads like behavior instead of planner metadata.
  • The Autopilot console now collapses into a simple handoff model: what the owner still does, what the rider takes over, and what to watch in the arena once DRAW hits.
  • House bots now steer around frontier anchors and converge toward stronger fight lanes instead of drifting into dead edges.
  • The planner now exposes an economy readiness score and confidence band so players can see when an agent is actually prepared to push a paid run.
  • The planner now also exposes objective posture so each doctrine explains whether it wants to contest, flank, or hold live arena objectives.
  • A Campaign Ops Queue turns planner output into the next owner-approved actions so players can execute an agent’s skill buy, paid run, or premium unlock in sequence.
  • Every agent now carries a persistent campaign ledger with wins, placements, treasury earnings, hot streaks, and a campaign tier.
  • A Frontier Tape panel records recent finished runs with placements, kills, score, payout, and settlement proof so the campaign feels like a real arc instead of a single match.
  • A live Autonomy Wire streams in-match directives so the player can see what autonomous riders are trying to do in real time.
  • The planner can drive one-click approval flows for the next recommended upgrade or paid run while keeping owner-signed X Layer actions honest.
  • Signed-in players can spectate live frontier matches and inspect ring state, paid pots, and the autonomy mix inside each showdown.
  • Live Frontier now works as a public board: matches can be watched without signing in, while signed-in players still get socket-backed live updates and leader cam.
  • The observer lane now supports live spotlight cards and leader cam so spectators can jump into the most active frontier round and follow the current leader.
  • Live Frontier cards now surface per-rider win history, streaks, tier labels, treasury linkage, premium state, and latest onchain motion so every live round has real rivalry context instead of anonymous dots.
  • The Live Frontier board now exposes quick filters and board-wide stats, so it is easy to scan paid vs practice activity and see how much of the active field is actually onchain-linked.
  • The Live Frontier board now includes a recent winners rail, so closed rounds keep feeding social proof back into the public product surface even after the live match has ended.
  • The Live Frontier board now adds a public leaders rail and a chain pulse feed, so viewers can see which riders are hot and which receipts are moving the frontier economy without digging into private tabs.
  • A Heat Check rail now calls out the riders winning the last cluster of frontier rounds, giving the public board a stronger rivalry and “one more run” pull.
  • Every public rider can now open a dossier with recent runs, recent receipts, streak context, treasury linkage, and a live spectate hook if that rider is currently in the field.
  • Live match cards are now flatter and faster to scan: the dossier keeps the deep stats off-demand while the board itself stays readable during spectating.
  • Field Intel now separates critical calls from the raw event feed, making eliminations, ring shifts, objective claims, and settlements easier to parse during live play.
  • Premium autonomy activations now feed back into the ledger as receipts, unlock expiry-aware planner guidance, and surface a structured x402 payment challenge in the UI.
  • The premium/x402 lane now shows a readable challenge quote, benefit checklist, and active-state summary so it feels like a real product lane instead of a raw payment payload.
  • The premium lane is now wired through a real browser-side x402 payment client, with the server issuing a standards-shaped payment challenge and bridging the signed payload into OKX's signed x402 facilitator endpoints.
  • X Layer skill purchase and match-entry flows.
  • Onchain settlement receipts stored and surfaced in the UI.
  • A Chain Ops Board summarizes registrations, skill buys, paid entries, settlements, premium activations, and the latest confirmed explorer link per agent.
  • The rider and chain consoles now explain where each agent sits in the campaign loop and onchain loop, so the product reads as a compounding system rather than disconnected widgets.
  • The onchain console now surfaces treasury linkage, career payout growth, and premium-lane state directly, making the OnchainOS wallet story legible instead of hiding it behind raw receipts alone.
  • The onchain console now includes a clear proof ladder for register → upgrade → paid entry → settlement, plus a public chain pulse so the selected rider’s private receipts sit in context with the wider frontier economy.
  • The rider loop now includes a visible Bounty Trail so each agent always has a next prize to chase: first finish, first settlement, hot streak, premium autonomy, or compounding treasury cycle.
  • OnchainOS wallet-account binding for agent treasuries.
  • x402 payment challenge route for premium autonomy passes.

Agentic Loops

  • Combat loop: autonomous agents decide when to chase, reload, dodge, rotate into the safe zone, and contest pickups.
  • Objective loop: signal-drop objectives pull riders into contested territory and reward whoever secures the drop with score, ammo, and healing.
  • Bounty loop: the arena periodically posts a live mark on one rider, forcing the field to hunt the leader or helping the marked rider bait bad pushes for a comeback swing.
  • Frontier event loop: roving stagecoach runs move through the town and give both human riders and bots a readable reason to collapse onto a live route instead of idling between fights.
  • Doctrine loop: each agent derives a doctrine from its skills, and the fallback combat brain now changes firing range, pickup routing, flanking, and center-control behavior to match it.
  • Positioning loop: saloon, hotel, wash, stable, and street cover now create real tactical lanes so autonomous agents can kite, reload, or bait bounty pushes from protection instead of wandering in open ground.
  • Terrain loop: solid landmarks and map-specific obstacles now physically shape paths, dodge lines, pickup routes, and bot rotations.
  • Progression loop: the Autonomy Director recommends the next highest-leverage skill buy based on the agent's current stat profile and receipt history.
  • Visibility loop: live autonomy directives explain why agents rotate, reload, contest supplies, or force a fight.
  • Economy loop: paid match entry, skill upgrades, and settlement all settle on X Layer, while the UI keeps showing the next onchain move the agent wants to make.
  • Budget loop: each rider can run inside a visible spend policy, so the player can decide how much upgrade budget to delegate without losing track of what the agent is allowed to do next.
  • Treasury loop: every agent is created with a linked treasury/subwallet track, so settlement outcomes can feed the next upgrade or queue decision.
  • Premium loop: the x402 autonomy pass is the premium lane for stronger planning, tighter queue discipline, and future higher-trust autonomous economy actions.
  • Premium state loop: when the autonomy pass is active, the planner switches posture, shows expiry, and records the premium activation as an onchain/autonomy receipt in history.
  • Campaign loop: finished matches roll into a long-lived career ledger so agents can build momentum, streaks, and treasury history across multiple showdowns.

x402 Payment Structure

  • The premium autonomy lane is intentionally modeled as a staged x402 flow instead of a one-off toggle.
  • The server returns a real 402 Payment Required response from POST /payments/x402/autonomy-pass when the agent has not yet settled the premium lane.
  • The browser now uses an x402-capable fetch client to sign the premium quote and retry the request automatically.
  • The premium quote is intentionally on X Layer mainnet, while gameplay skill buys, paid entry, and settlement remain on X Layer testnet.
  • The default repo config points the x402 lane at USD Coin on X Layer mainnet with asset address 0x74b7f16337b8972027f6196a17a631ac6de26d22 and a default quote amount of 1000000 base units.
  • The web surfaces that as a structured payment challenge showing amount, asset, chain, recipient, and the current premium-lane checklist.
  • Once the payment settles through the configured OKX/OnchainOS flow, the app:
    • creates an autonomy_pass receipt,
    • stores the expiry window,
    • flips the planner into premium mode,
    • and shows the pass inside the same agent ledger as skill buys, match entries, and settlements.
  • This keeps the premium AI loop honest: players can see exactly when the premium lane was activated, what it unlocked, and how it feeds the agent’s economy routing.
  • The remaining gap is funding: the code path is live, but a public mainnet x402 proof hash still needs a funded USDC wallet to exercise the full lane end to end.

Honest Autonomy Model

  • Today, player-owned agents can autonomously fight and plan, but onchain skill purchases and paid entries still require the player wallet signature because the contract enforces owner-signed actions.
  • The new budget-autopilot lane respects that contract model: it can decide when a buy fits the cap, prompt the next owner-signed skill purchase, track how much delegated budget has been consumed, and auto-stage follow-up practice runs without pretending the contract is more autonomous than it really is.
  • House bots are fully operator-managed and can register and enter matches without user intervention.
  • The current product therefore supports agent-directed onchain actions with user approval, not invisible custodial spending for player-owned agents.
  • This is intentional: it keeps the X Layer proof real while preserving a clear path toward deeper OnchainOS-managed autonomy.

Quick Start

  1. Install dependencies:
pnpm install
  1. Copy the env template and fill the values you have:
cp .env.example .env

At minimum, fill the X Layer testnet values, NEXT_PUBLIC_ARENA_ECONOMY_ADDRESS, the operator key / treasury pair, and the OnchainOS + OKX payment credentials if you want to exercise the premium x402 lane.

  1. Start Postgres.
docker compose up -d postgres
  1. Run a testnet preflight before deploying.
pnpm --filter @rdr/contracts preflight:testnet
  1. Deploy the contract to X Layer testnet after setting ARENA_OPERATOR_PRIVATE_KEY and APP_TREASURY_ADDRESS.
pnpm --filter @rdr/contracts deploy:testnet

The deploy command writes packages/contracts/deployments/xlayerTestnet.json and prints the env lines to copy.

  1. Set NEXT_PUBLIC_ARENA_ECONOMY_ADDRESS in .env to the deployed contract address.

  2. Start the app stack.

pnpm dev

The web app defaults to http://localhost:3000 and the server defaults to http://localhost:4000.

Build And Test

pnpm build
pnpm test

OnchainOS Notes

  • Agent wallets are generated locally and optionally bound to an OnchainOS wallet account through the Wallet API when the OKX API credentials are configured.
  • The x402 route is exposed at POST /payments/x402/autonomy-pass.
  • The x402 browser flow now expects X Layer mainnet wallet access plus a supported asset balance, while the game economy continues to use X Layer testnet.
  • The autonomy planner endpoint is exposed at GET /agents/:id/autonomy-plan.
  • The campaign ledger endpoint is exposed at GET /agents/:id/campaign.
  • Recent finished runs are exposed at GET /agents/:id/matches.
  • The product is structured so x402 is not just a payment stub; it is the premium autonomy lane for higher-trust planning and future agent economy automation.
  • The current implementation uses the OKX /api/v6/x402/supported, /api/v6/x402/verify, and /api/v6/x402/settle endpoints with signed headers when payment payloads are supplied.
  • ONCHAIN_OS_WALLET_BASE_URL and OKX_PAYMENTS_BASE_URL must be root hosts such as https://web3.okx.com, not full /api/... paths.

X Layer Notes

  • The repo currently defaults to the recent X Layer testnet configuration: chain ID 1952, RPC https://testrpc.xlayer.tech/terigon, explorer https://www.okx.com/web3/explorer/xlayer-test.
  • The premium x402 lane defaults to X Layer mainnet: chain ID 196, RPC https://rpc.xlayer.tech, explorer https://www.okx.com/web3/explorer/xlayer.
  • OKX has published inconsistent testnet snippets on different pages. If your wallet or RPC provider expects a legacy config, override the env values instead of editing code.
  • The web wallet config now also reads NEXT_PUBLIC_XLAYER_TESTNET_CHAIN_ID, so the browser and server can be kept aligned if the live network uses a different testnet chain ID.

Live Deployment

  • ArenaEconomy address: 0x31a44d5dcA53A0BFB13C79d8dF5ED3148f08DB97
  • Deployment tx: 0xf6573f85ca2dfdc1e4cfee1a027782a1c620d918e3ce984280c12dacb268386a
  • Deployer: 0x1620e0aEd0215Add62EF51A109De35455D35B6A1
  • App treasury: 0x1620e0aEd0215Add62EF51A109De35455D35B6A1
  • Network: xlayerTestnet
  • Chain ID: 1952
  • RPC: https://testrpc.xlayer.tech/terigon
  • Explorer: OKX X Layer Testnet Explorer
  • Deployment block: 25006326
  • Deployed at: 2026-03-14T05:12:43.260Z
  • Deployment artifact: packages/contracts/deployments/xlayerTestnet.json

Submission Proof

Confirmed live X Layer testnet proof captured so far:

The proof checklist, explorer links, and reproduction notes live in docs/proof.md.

The live deployment checklist lives in docs/testnet-runbook.md.

About

Agentic western arena game on X Layer with OnchainOS

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages