← Back to Home

The static grid from my last update has come alive.

When I last wrote about the Gamma Project, we had a board and some random balls. It was a start, but it wasn’t a game. Today, I’m happy to report that the core soul of “Lines”—the mechanics that keep you clicking for “just one more turn”—is fully implemented.

The Mechanics of Flow

Turning a grid of colored circles into a playable game required solving three distinct logic puzzles: how to move, how to find matches, and how to reward the player.

1. Finding the Way (BFS)

In Lines, balls don’t just teleport. They travel. And they can’t jump over other balls. To implement this, we used the Breadth-First Search (BFS) algorithm.

When you click a ball and then an empty square, the game calculates if a clear path of empty cells connects them. If the path is blocked by other colors, the move is forbidden. It adds that crucial layer of strategy: you’re not just matching colors, you’re managing space.

2. The Satisfaction of the Line

The heart of the game is the pop. Aligning 5 balls of the same color should feel instantly rewarding.

We implemented a robust line detection algorithm that scans the board in four directions—Horizontal, Vertical, and both Diagonals—after every move.

If it finds a sequence of 5 or more matching balls:

1. They vanish. The board clears up, giving you breathing room.
2. You get a free turn. No new balls spawn, allowing you to chain combos.

3. The Scoreboard

We finally have a way to keep score! Sticking to the classic rules:

• 10 points for a standard line of 5.
• +2 points for every extra ball in that line.

It differentiates a lucky save from a masterful 7-ball cross-clear.

The Agentic Workflow: Coding with a Partner

What makes this project unique isn’t just the code, but how it’s being written. I’m working with Google Antigravity, an AI agent.

For the scoring system, our workflow looked like this:

1. Planning: I asked for scoring. The agent researched standard “Lines 98” rules to ensure authenticity.
2. Implementation: It drafted the code to track state and update the DOM.
3. Verification: This is the magic part. The agent didn’t just say “done.” It spawned a browser instance, played the game itself, placed 5 balls, checked the score, placed 6 balls, checked the math, and proved it worked before handing it back to me.

It feels less like using a tool and more like pair programming with a tireless engineer who loves writing test cases.

What’s Next?

The game is playable! You can move, score, and lose (when the board fills up).

Next, we need to polish the rough edges:

• Preview bubbles: Showing where the next colors will land.
• Animations: Making the movement and clearing feel smooth and impactful.

Play the latest build here and let me know your high score!

← Back to Home