Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

60 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OrbitForge

OrbitForge is a hackathon demo for a seeded, self-improving AI operations stack around space-based datacenter infrastructure.

Public repo: https://github.com/markmdev/orbitforge

Current Demo

The app is a runnable orbital-compute mission console. It shows:

  • seeded orbital nodes, ground stations, workload scenarios, and link/thermal constraints;
  • Gemini 3.5 operator plan and improvement critique traces through app runtime API routes, with recent learning memory included in plan/critique/audit context;
  • deterministic scorecards, policy mutation, A/B sweep, learning-memory ledger persisted in browser state, explicit operator-run improvement pass, scenario-scoped promotion gate, first-viewport next-action command strip, scenario-aware incident command deck, stateful operator promotion, seeded mission execution timeline, delivered data-product manifest, actionable incident work queue, operations log, and a visible unsafe-policy guardrail canary;
  • seeded stress-drill generation that appends a new what-if scenario to the current evaluation library and gives it a command deck;
  • Gemini 3.5 Flash computer-use audit path with a generated audit frame, propose-only actions, prompt-injection guard state, and exact quota/API blocker display;
  • judge report export with active policy state, current scores, incident readiness, applied commands, mission execution state, Gemini runtime health, Gemini status, audit mode, prompt guard, and seeded-data guardrail.

Known live state: Gemini API quota is currently returning You do not have enough quota to make this request. The app preserves the exact blocker and falls back honestly instead of presenting fallback output as live.

Run Locally

npm install
npm run dev

The dev server runs at http://127.0.0.1:5174/, matching the runtime verifier.

Useful proof commands:

npm test
npm run build
npm run verify:runtime
npm run verify:preview
npm run verify:demo

Use npm run verify:gemini only when a live Gemini quota check is worth spending a model call.

GitHub Actions runs npm run verify:demo on push and PR with a placeholder GEMINI_API_KEY, proving configured-state middleware without making live Gemini calls.

Project Map

  • HACKATHON.md for theme and prize strategy.
  • REQUIREMENTS.md for current product requirements.
  • ARCHITECTURE.md for the initial implementation shape.
  • QUALITY_BAR.md for the product, research, design, AI, and engineering bar.
  • IMPLEMENTATION_PLAN.md for the build sequence.
  • RUNBOOK_24H.md for the planned 24-hour execution model.
  • LINEAR_PLAN.md for the Linear tracking structure.
  • THREAD_ORCHESTRATION.md for specialist Codex thread roles.
  • EVAL_PLAN.md and QA_MATRIX.md for proof strategy.
  • TOOLS_AND_ACCESS.md for account/tool prerequisites.
  • SEED_DATA_PLAN.md for the first scenario/entity set.
  • DEMO_SCRIPT.md for the judge-facing narrative.
  • ADRs/ADR-0001-gemini-self-improving-orbital-ops.md for the accepted product direction.
  • PRDs/PRD-0001-orbitforge-hackathon-demo.md for the first build spec.
  • knowledge-base/README.md for the linked research map.
  • reports/ for synthesis reports.
  • notes/ for durable work memory and follow-ups.

About

Gemini 3.5 self-improving orbital ops hackathon demo

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages