How to use Codex to synchronize a shifting labyrinth
Codex helped synchronize Wizard’s War’s shifting maze with authoritative ticks and shared physics; a playable beta still needed broader real-player testing.
devpost.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Audit our shifting labyrinth’s server and client state contract. Swap maze generation and collision geometry at one authoritative tick, preserve sequenced-input prediction and reconciliation, and add native/WASM parity cases. Report synthetic load separately from real-player testing and present the playable change for human review.
Reach people exploring AI agents
See placements →What happened
A playable beta coordinates shifting labyrinth state with shared physics and server-authoritative inputs; larger production playtesting remained future work.
Reported by Christopher Peret — DevpostHow the agent did it
- Peret preserved the earlier alpha and directed Codex’s beta implementation and performance boundaries.
- Codex coordinated labyrinth swaps with authoritative ticks and shared client/server collision geometry.
- The project exercised prediction, reconciliation and synthetic load behavior, while the creator inspected the browser experience.
Where a person stepped in
Peret retained rules, balancing and performance decisions and tested the playable experience; synthetic bots do not establish production player capacity.
Christopher Peret used Codex to extend Wizard’s War from a preserved earlier alpha into a multiplayer beta. A central challenge was the shifting labyrinth: changing a maze is not just a visual update when clients and the authoritative server share collisions and movement.
The implementation uses a C++ server and a TypeScript/Three.js client with shared native and WebAssembly physics. Clients submit sequenced inputs rather than authoritative positions or damage. Codex helped synchronize labyrinth changes at the same authoritative tick, keeping collision geometry and state contracts aligned across both sides.
The account also describes client prediction, server reconciliation and checks that compare the shared geometry. Human direction supplied the game rules, balancing and performance constraints, with browser testing needed to judge the playable experience. Earlier work remained identified as the baseline rather than being attributed entirely to the event.
A reported cast-storm test with hundreds of bots was a synthetic load result, not proof of that many concurrent human players. The concrete artifact is a playable beta and an implementation that coordinates moving-world state. Larger production playtesting and the broader persistent-world ambition remained further work.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Devpost · Christopher Peret
Frequently asked questions
What may clients submit?
Sequenced inputs, rather than trusted positions or damage outcomes.
Why synchronize maze changes?
Rendering and collision state must agree across server and clients.
Did the bot test prove human-player capacity?
No. It was a synthetic stress result.
More details
About this account
Peret preserved the earlier alpha and directed Codex’s beta implementation and performance boundaries. Codex coordinated labyrinth swaps with authoritative ticks and shared client/server collision geometry.
Original dated creator submission opened. Codex’s beta work is separated from the preserved earlier alpha. Synthetic bot performance is not a production-capacity guarantee, and no independent game test was performed. Suggested prompt written by Agents In Use; no original task prompt was found in the verified text.
Credit: Christopher Peret — Devpost
Source published 21 Jul 2026
Source checked 14 Oct 2026
Tools and topics
Codex · C++20 · TypeScript · Three.js · WebAssembly · Box3D
multiplayer simulation · authoritative state · shared physics
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.