# Copy-ready NASA Space Apps submission — model 1.1.0

Existing team: gunnchOS, Chicago. Registered challenge: Build a Junior Astronaut Mission Trainer. Existing baseline submission was confirmed November 15 at 4:09 AM America/Chicago. This document prepares the strengthened revision; final resubmission status is recorded in the release handoff after portal confirmation.

## Project title

7GC Mission Control — Lunar Outpost Resilience Trainer

## High-level summary

Keep two astronauts safe when their lunar outpost loses power. 7GC Mission Control turns NASA-based oxygen, water and food budgets into a playable engineering lesson. Learners repair a solar-array fault, shelter during a radiation drill and choose between science, greenhouse care and reserves. A deterministic planner compares 567 phase policies and exposes what each sacrifices. Junior, Explorer and Engineer views share one simulation. The mission, planner and explanations work offline, with exportable traces and explicit scientific limitations.

## Final project URL

https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/

## Official project demonstration URL — seven slides

https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/7gc-mission-control-slides.pdf

Editable version: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/7gc-mission-control-slides.pptx

The portal’s demonstration field requires a seven-slide presentation or a video of at most 30 seconds. Use the PDF there. The separate interactive judge walkthrough is linked in the detailed description below.

## Detailed project description

7GC Mission Control is an interactive engineering lesson for the Build a Junior Astronaut Mission Trainer challenge. The player directs a fictional lunar outpost with two crew during a 24-hour exercise. Their decisions change actual electrical demand, oxygen generation, water recovery, dry-food stocks, scientific work, greenhouse demand and simulated science-data delivery.

An array circuit fault at hour 6 reduces solar supply. A robotic repair adds short-term power demand and restores generation only after its powered repair intervals finish. A radiation drill at hours 10–14 requires shelter, which adds demand and pauses science. A simulated radio outage at hours 12–18 buffers data locally; an enabled radio later drains the queue, while a full buffer explicitly drops new data. All events and hardware parameters are authored teaching assumptions.

The core scientific inputs come from NASA NTRS 20230012635: 0.84 kg oxygen/person/day for the 82 kg surface reference crew, water-use components totaling 3.20 kg/person/day, and 0.98 kg dry food/person/day for the non-EVA reference case. Electrolysis consumes approximately 1.125 kg water per kg oxygen generated. NASA’s ECLSS overview explains the water/oxygen coupling and supplies a historical recovery benchmark. The model processes only available wastewater and feedwater. Resource traces distinguish actual served mass, demand, unmet demand, treatment reject, collection losses, overflow and vented hydrogen.

The preferred technical contribution is comparative planning. One deterministic state transition drives the game, forecast, planner, sensitivity runs and exported replay. The planner exhaustively evaluates 567 phase-policy specifications from identical initial resources: alternative repair times, science windows and greenhouse-care targets. It checks both mission completion and explicitly authored reserve margins. In reference conditions, 199 policies meet those margins, and five nondominated science/crop-care choices show the competing objectives. The planner does not claim a global optimum over arbitrary hourly decisions.

Verified reference choices include a science-priority plan with 12 science hours and no greenhouse care, a balanced plan with 9 science hours and 12 crop-care hours, and a greenhouse-priority plan with 4 science hours and 24 crop-care hours. The balanced plan retains a minimum battery reserve of 0.75 kWh and completes with 1.20 kWh energy, 1.53 kg oxygen, 3.52 kg water and about 1.04 kg dry food. Crop-care hours measure supplied activity, not edible yield, crop health or proven future harvest.

Sensitivity reruns fixed decisions with 10% less solar power, 10% more crew consumption, an extra repair hour and all three changes together. Eighteen policies satisfy the same reserve requirements in all five tested conditions. A selected robust policy sacrifices scientific work and greenhouse care—2 and 4 hours respectively—to retain at least 1.035 kWh in the combined case, without retuning. This is bounded deterministic evidence, not a probability of reliability.

Junior mode offers friendly language, a mission dictionary, readable controls and user-paced time. Explorer adds what-if conditions and policy comparison. Engineer exposes equations, ledgers and hourly traces. Switching view leaves the mission state unchanged. The intended learning task is to explain how one resource-saving decision changes another system and why a preferred plan gives up an activity. We have not measured classroom outcomes or obtained accessibility certification.

The runtime is dependency-free HTML, CSS, JavaScript and original code-native SVG. It requires no account, API key or remote inference. The self-contained offline HTML includes the mission, planner and science explanations. A service worker additionally supports offline reload after first load. Forty-one model/planning tests cover nominal and failed missions, physical/queue accounting, state validation, policy feasibility, nondominance, sensitivity and exact replay. Real-browser checks exercise manual controls, the applied robust plan, learning-view invariance, glossary, keyboard map, mobile bounds, export and both offline paths. Source code and machine-readable results are publicly available.

Limitations are explicit: no cabin pressure or CO₂ dynamics, thermal balance, battery efficiency/rate limits, crop biology, real radiation transport, full habitat mass closure or physically derived RF link. Food-stock exhaustion is a mission logistics rule, not a prediction of immediate medical harm. Exposure points are illustrative, not clinical dose. No live data, NASA certification or endorsement is claimed. Next work should begin with educator/learner and assistive-technology testing, then more detailed physical models and constrained scheduling.

Project lead: Edmund Gunn Jr., gunnchOS. OpenAI Codex generated and revised the implementation, illustrative parameters, explanatory text, original SVG design, tests and presentation under the owner’s project direction. The gunnchOS/7GC names and lunar-trainer concept predate this implementation. All new work described here was authored November 15, 2026; no November 14 coding work is claimed. Existing production repositories, game/character assets, source leases, release gates and active worktrees were not reused or changed.

Tools used: Node.js standard library/test runner; Python for local serving and packaging; Playwright and Google Chrome for browser checks; OpenAI Sites for isolated hosting; OpenAI presentation artifact tools for editable slides; ReportLab and pypdf for the portable rendered PDF and structural checks. Original AI-assisted graphics and code have visible/descriptive/metadata disclosures. The deployed game contains no third-party runtime library. Third-party tools retain their own licenses; original submitted solution content follows the default Apache 2.0 specified by the accepted 2026 participant terms. No Azure service, paid inference, NASA logo, stock photo or unlicensed third-party artwork was added.

Interactive judge walkthrough: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/demo
Technical model and assumptions: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/MODEL.md
All computed policy results: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/planning-evidence.json
Judging evidence map: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/EVIDENCE.md
Reproducible source: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/7gc-mission-control-source.zip
Standalone offline app: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/offline.html
Four-minute pitch and Q&A: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/PITCH.md

## NASA Data field

- NASA NTRS 20230012635 — Tables 2, 8 and 11: https://ntrs.nasa.gov/citations/20230012635
- NASA Environmental Control and Life Support Systems: https://www.nasa.gov/reference/environmental-control-and-life-support-systems-eclss/
- NASA Space Apps 2026 — Build a Junior Astronaut Mission Trainer: https://www.spaceappschallenge.org/2026/challenges/junior-astronaut-mission-trainer/

## Space Agency Partner & Other Data field

No other space-agency dataset is claimed. Cite the development/verification tools and original source package:

- OpenAI Codex — AI-assisted code, text, research and original SVG design: https://openai.com/codex/
- OpenAI — Sites hosting and presentation artifact tools: https://openai.com/
- Node.js — build and deterministic tests: https://nodejs.org/
- Python — serving and packaging: https://www.python.org/
- Playwright — browser verification: https://playwright.dev/
- Google Chrome — browser verification: https://www.google.com/chrome/
- ReportLab — rendered PDF assembly: https://www.reportlab.com/opensource/
- pypdf — PDF checks: https://pypdf.readthedocs.io/
- Original source and offline kit: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/7gc-mission-control-source.zip
- Authorship and license notes: https://gunnchos-7gc-lunar-mission-2026.tenaciousfjord2.chatgpt.site/THIRD_PARTY.md

## AI-use selections

Ideation or Brainstorming; Research or Information Gathering; Data Analysis and Interpretation; Coding, Debugging, or Troubleshooting; Writing, Editing, or Translation; Visual Design or Image Creation; Presentation, Video, or Audio.

Do not select “Building SI/ML Into the Project”: the runtime contains no AI inference. No generated audio or video is claimed.

## Submission operator notes

Use the existing gunnchOS Project page. Preserve team, challenge and event. The official 2026 guide permits edits and resubmission before the close; only the last submitted revision is reviewed. Save Progress first, verify saved content, then Submit for Judging and retain the actual banner and Last Submitted timestamp. A saved draft alone does not secure judging eligibility.

Chicago owner-provided cutoff: 1:30 PM America/Chicago on Sunday November 15. Global cutoff: 11:59 PM local time. Owner reports no separate local link is available. Do not claim a separate local form was submitted.