How to use Gemini to build a game on a phone
Gemini helped Lakshman build a Rune game in Termux on his phone; complete logic files resolved AI-player and departure handling before multiplayer testing.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Build this Rune chain-reaction game from Gemini CLI in Termux on my phone. Implement cell thresholds and neighboring nanobot propagation, then use the complete reference logic file for the AI player and player departure behavior. Preserve work in Git and pause for my phone playtests of joining, leaving, ownership cleanup, and multiplayer rounds.
Reach people exploring AI agents
See placements →What happened
The account links a working Ripple Effect game on Rune and its repository, reporting up to six players after phone-based development and testing.
Reported by Lakshman on HashnodeHow the agent did it
- Lakshman configured Termux, Node.js, Git and Gemini CLI on his Pixel phone.
- Gemini implemented chain reactions from the author's behavior description.
- Full reference files resolved AI-player and departure handling before phone-based testing.
Where a person stepped in
Lakshman defined chain-reaction behavior, supplied complete reference files when smaller snippets failed and tested and corrected the game on his phone.
Lakshman built Ripple Effect for Rune using Gemini CLI inside Termux on a Pixel 7. His phone environment included Node.js and Git, and development happened during metro and bus commutes rather than at a desktop.
The game adapted chain-reaction board behavior: selecting a cell adds nanobots, and crossing its threshold sends them into neighboring cells. Lakshman explained that core logic to Gemini, which generated it successfully in the first reported attempt. Git preserved the evolving files while he continued giving direct instructions.
Two later features needed better context. A random-cell AI player remained broken through several prompts and partial examples. Supplying the full logic.ts file from an earlier project finally made the desired behavior clear. Player join and leave handling required the same approach, including removal of a departing player's nanobots.
After seven days, Lakshman reports a working game supporting up to six players and links both the playable Rune version and its repository. He tested it and fixed bugs on the phone. The concrete lesson in the account is that complete relevant files resolved failures that isolated snippets and repeated instructions did not; the agent still depended on his game rules and hands-on play testing.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hashnode · Lakshman
Frequently asked questions
What ran on the phone?
Termux hosted Node.js, Git and Gemini CLI on a Pixel 7.
What fixed the AI-player implementation?
The author supplied the entire logic.ts file from an earlier project.
Was there a playable result?
The account links Ripple Effect on Rune and reports support for six players.
More details
About this account
The game adapted chain-reaction board behavior: selecting a cell adds nanobots, and crossing its threshold sends them into neighboring cells. Lakshman explained that core logic to Gemini, which generated it successfully in the first reported attempt. Git preserved the evolving files while he continued giving direct instructions.
Primary firsthand account opened with linked game and repository. Seven-day development and six-player support are reported by author, not independently tested. ChatGPT helped the initial prompt; Gemini CLI performed file editing. Prompt is suggested by Agents In Use, not a quoted original.
Credit: Lakshman on Hashnode
Source published 29 Jul 2025
Source checked 14 Oct 2026
Tools and topics
Gemini CLI · Termux · Node.js · Git · Rune
Mobile development · Multiplayer game
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.