[Photon] Agronaut

• 

  Project Cover

• 

  Introduction

• 

  Operation UI - Overall

• 

  Operation UI - Greenhouse

• 

  Conversation1 with Photon as Earthside Liaison

• 

  Conversation2 with Photon as Earthside Liaison

• 

  Ranking

• 

  Player Parttern

• 

  Report

• 

  Report Analysis

Beyond the Flight

“Fly Me to the Moon” made us ask what happens after landing: how would people grow food and sustain a lunar base? Agronaut turns that question into a playable mission where people build lunar farms, face hazards, and test their own strategies.

Inspiration

If people are going to live beyond Earth, growing food cannot be an afterthought. A lunar farm would have to share limited space, water, and power with the systems keeping its crew alive. We found those connections fascinating, but they are difficult to grasp from a diagram alone. What happens to a greenhouse when its power route is damaged? Is a highly productive layout still a good one if it cannot survive a crisis?

We wanted to let people explore those questions by making decisions themselves. In a game, placing a water recycler, connecting a corridor, or choosing what to repair can have consequences several turns later. That makes the science more tangible and gives players a reason to experiment.

We also saw an opportunity beyond education. Every player can approach the same constraints with a different layout and strategy. By preserving those decisions and comparing their outcomes, we could build a collection of ideas that sparks new questions for space agriculture. That combination of public learning and collective exploration became Agronaut.

What It Does

In Agronaut, players manage a lunar agriculture base from its initial design to its final mission assessment. They begin by planning the layout and connecting essential systems, then operate the base as it grows and conditions change. Players can choose a Challenge or a longer Progressive mission. Hazards can disrupt their plans, so a strong farm needs both productive systems and a strategy for keeping them running.

Much of the base-building design draws on NASA research. The game brings ideas from controlled-environment agriculture and closed-loop life support into decisions players can see and test. Water, power, oxygen, temperature, and food are connected: changing one part of the base can affect everything else. As the mission unfolds, players see the consequences of earlier choices and adapt their approach. They learn those relationships by experimenting with layouts and seeing the results unfold.

To make the mission more immersive, players can optionally contact Photon, an AI Earth liaison, through iMessage. Photon will be always connected to the progress of your current mission, offering limited hints without choosing for the player and sends warnings when trouble reaches the lunar base. The conversation unfolds alongside the mission, making Earth feel present even as players make their own decisions on the Moon.

After each mission, players receive a report on their decisions and the outcome. They can replay what happened, compare approaches through rankings for production, stability, efficiency, resilience, and overall performance, and even view top ranks' real layouts and logs. Player Patterns brings these individual missions together, revealing broader trends in how players design and operate their bases. These aggregated patterns can also serve as a reference for space agriculture researchers exploring future layouts and strategies.

How We Built It

We built Agronaut with Next.js, React, TypeScript, and Phaser 3. Its simulation models the base as a connected network, then resolves agricultural growth, resource use, hazards, and player actions turn by turn. Supabase stores completed missions, reports, replays, and rankings.

To keep the leaderboard tied to actual gameplay, the server receives a player's action transcript and replays the entire mission before calculating and saving its score. Run IDs are bound to their transcripts, submissions have rate limits, and saved actions remain available for replay. Scores combine transparent game-rule measures with an AI strategy assessment.

We use curated NASA and ESA material to give the agricultural and lunar setting scientific context. OpenAI supports strategy assessment and mission reports, while Photon Spectrum connects the optional iMessage conversation to the active game through signed webhooks.

Challenges We Ran Into

Making layout matter took repeated balancing. A greenhouse's position alone is not enough: its connection to the utility network, nearby water recycling, available power, and exposure to hazards all affect the mission. We wanted those relationships to feel intuitive and consequential in play.

Keeping rankings trustworthy required us to validate entire missions, not just final numbers. We built server-side replay so saved scores correspond to legal construction and actions.

Making Photon feel like Mission Control was both a writing and integration challenge. It needed to stay in character, give useful hints without solving the game, and respect the two-question limit. Getting a real iMessage opening, player reply, webhook, hint, and disaster warning working end to end also required careful configuration of message routing and webhook signatures.

Accomplishments That We're Proud Of

We completed the loop from building a base to operating it, receiving a report, appearing in Records, and replaying the mission. We also verified a live iMessage conversation with Photon: the player received the opening, got two hints, encountered the weak-signal limit, and later received a disaster warning.

We're proud that a saved result retains the decisions behind it. Someone looking at a score can inspect the layout and follow the turns that produced it.

What We Learned

A final score says little about why a strategy worked. Recording the layout, accepted actions, resource changes, and hazards made our reports and comparisons more useful.

We also learned how much the experience depends on connecting scientific ideas to clear player choices. When a change to the base has a visible effect several turns later, players can begin to understand the trade-offs—and come up with strategies we would not have designed ourselves.

What's Next for Agronaut

We want to make it easier to compare genuinely different player layouts and identify recurring strategies across more missions. Longer term, we hope to work with space agriculture researchers to review promising design questions, improve the simulation's assumptions, and turn the most interesting player ideas into hypotheses for more rigorous testing.

Built With

• css
• javascript
• next.js
• phaser.js
• photon
• postgresql
• python
• react
• sql
• supabase
• typescript