Knuckle Rush - Generated Levels and Corner Picks Update

I am still working on Knuckle Rush for the Platform Game Jam. Here are the updates I added last night and today.

Last Night's Updates

• Added the new "Tristan Presents" opening cutscene.
• Connected Supabase so Online mode can use shared best times.
• Improved Normal and Online time records for each boxer and difficulty.
• Updated the ghost racer to follow the best saved time.
• Added different gyms and backgrounds between stages.
• Improved gym lighting, ropes, platforms, signs, and surroundings.
• Added stronger enemy types and bosses.
• Improved boxer movement, combat, health, guard, combos, sounds, and camera movement.
• Fixed the game so players must defeat enemies before moving to the next platform.

Today's Updates

• Levels now continue generating after the designed opening stages.
• Every difficulty has its own boss schedule:
  • Rookie: boss every 4 stages.
  • Pro: boss every 3 stages.
  • Champion: boss every 2 stages.
• Normal and Online modes now use the same level and boss progression.
• Gyms change based on the selected difficulty.
• Added automatic checks to keep platforms connected and playable.
• Enemy positions are checked so they stay on their platforms.
• The bell is kept safely on the final platform.
• Added the designed Corner Cut stage as Stage 2.
• Improved the Next Stage flow.
• The next stage stays locked until a corner upgrade is selected.
• Corner upgrades now carry into the next stage correctly.
• Fixed corner picks so they cannot activate twice or use an old selection.
• Checked 180 generated stage and difficulty combinations.
• Tested all five corner upgrades.

Current State

The game is playable in Normal and Online modes with Rookie, Pro, and Champion difficulties. Players can race their best time, fight through generated stages, choose upgrades, face bosses, and continue their run.

I am still polishing the gameplay, balance, graphics, enemies, levels, and online features. Thanks for playing and testing Knuckle Rush.

Made by Tristan with AI development help from OpenAI Codex.