Honk Jam — Car Parking Puzzle
Unity 6

Honk Jam — Car Parking Puzzle

Unity Car Parking Puzzle with a Solver-Verified Level Pipeline

Overview

Honk Jam is a car parking puzzle about doing things in the right order: tap a car, it drives out of the lot, and passengers of the matching color board it at the pickup dock. It is the third Group Taiga game and the first one built in Unity 6 (C#, URP), running on the shared Go game backend. Every level is run through a solver before it ships, and the game contains no ads.

What It Does

Players clear 500 levels in five packs (the first bundled, the rest delivered from a CDN), meeting 11 mechanics such as buses, cones, hidden-color cars, key and lock cars, one-way barriers, hitched cars, garages, a second deck, timed levels and VIP riders. A daily puzzle has its own time leaderboard, a Classic Rush-Hour mode adds 160 levels, and boosters, daily tasks, a login streak and garage skins round out the loop.

Tech Stack

Unity 6C#URPRevenueCatGoPostgreSQLRedisDockerKubernetes

Key Features

500 levels in five packs with 11 mechanics, each introduced with an explanation card

Daily puzzle with a time-ranked leaderboard, plus weekly, country and global boards

Classic Rush-Hour mode with 160 BFS-solved levels

Undo, extra dock slot, queue shuffle and a hint that finds a safe next move

Daily tasks, piggy bank, login streak and garage skins

Offline outbox with optimistic updates; no ads

24 in-game languages

Gallery

Honk Jam — Car Parking Puzzle screenshot 1Honk Jam — Car Parking Puzzle screenshot 2Honk Jam — Car Parking Puzzle screenshot 3Honk Jam — Car Parking Puzzle screenshot 4Honk Jam — Car Parking Puzzle screenshot 5

Technical Highlights

→Unity 6 with URP, Input System and IL2CPP ARM64; an engine-independent C# rules engine under the presentation layer
→Rule-engine parity: 1,000 shared golden cases must produce identical results in Go and in Unity tests, both run in CI, with a versioned rule set
→Level pipeline: reverse-placement generator, memoized DFS solver with canonical states, measured difficulty and an 8-symmetry SHA-256 fingerprint for de-duplication; a ~7,000-level pool builds in about six minutes
→Difficulty tuned with a sawtooth curve and calibrated from player data with a logistic-regression tool
→Anti-cheat: the server strictly replays each attempt's tap, undo, slot and shuffle log, and daily times are bounded by server-witnessed time
→Go backend on the shared game base (chi, pgx, go-redis, MessagePack + zstd) with Redis leaderboards and A/B remote config, deployed via GitLab CI to Kubernetes
→Performance pass: iOS target raised from 30 to 60 fps, car meshes cut from 13-16k to 2.3-3.8k triangles, hint-solver allocations reduced by 69%

Key Achievements

01

Shipped the third Group Taiga game and the first Unity title to the App Store and Google Play

02

Kept two rule engines in two languages provably in sync with 1,000 golden test cases

03

Hardened the economy in a backend audit so signed reward events pay out exactly once under replay and concurrency

04

Built server-side ad verification for later use while shipping an ad-free game

Impact & Results

Honk Jam carried the shared game base from React Native into Unity, proving that the same Go backend, economy and anti-cheat model can serve a native-engine game with a much richer visual layer.