NOC Ninja — Guided NOC Patrol
Investigate three Arista EOS link incidents through twelve guided decisions. Compare recorded commands at both endpoints, choose the replacement sequence, and review the diagnosis or reported repair. Scenario, question and answer order shuffle each patrol. Drag the shuriken to answer, learn from feedback, and download a handoff.

The builder
- Name
- Cat Gurinsky
- From
- Austin, TX
- Org
- Immortal Martial Arts Academy
- First spec project?
- Yes
- Began at
- unknown
- Links
How it was built
- Model
- unknown
- Harness
- Codex
- Spec method
- GitHub Spec Kit 1.4.2.dev0
- Spec artifacts
- constitution, spec, clarify, plan, tasks, checklists, research, data-model, contracts, verification
Build environment
- MCP servers
- mcp__codex_apps
- mcp__cua_repl
- mcp__node_repl
- Sources & RAG
- StarterKit/CONSTITUTION.md
- StarterKit/SPEC.md
- StarterKit/PLAN.md
- Owner-supplied Arista EOS command captures and deployment confirmation
- Owner-supplied optics-reference.yaml (authoring reference; not a runtime classifier)
- Owner-defined technician replacement runbook
- Arista EOS Ethernet Ports and Transceiver Performance Monitoring manuals
- Arista 7050X3 and 7800R3 datasheets
- MDN JavaScript modules and Fetch documentation
- W3C accessibility guidance
- Owner-linked Adobe Stock ninja image as an inspected style reference only
- Skills
- speckit-constitution
- speckit-specify
- speckit-clarify
- speckit-plan
- speckit-tasks
- speckit-analyze
- speckit-implement
- imagegen
- unified-computer-use (native Safari verification; browser connector unavailable)
- Other tools
- Built-in image_gen
- Node.js 26.13.0 and dependency-free test harness
- Chrome Beta 159.0.8154.6 with local DevTools Protocol
- Python static HTTP server
- Git
- ripgrep
- curl
- Shell commands and file patches
- Web search and image inspection tools
- AI assets
- assets/fuji-maple.webp — original Mount Fuji and red maple quiz-page background generated with built-in image_gen
- assets/ninja.png — original transparent ninja sprite generated with built-in image_gen
- assets/shuriken.png — original transparent shuriken sprite generated with built-in image_gen; also used as the browser favicon
- assets/arcade-card.webp — rendered promotional composition using the existing generated sprites
- assets/himeji-castle.webp — original Himeji Castle-inspired twilight arena background generated with built-in image_gen
Runtime: how to play
- Type
- static-web · HTML5, CSS, JavaScript ES modules, Authored JSON, Local PNG/WebP assets
- Start
- python3 -m http.server 8001 --bind 127.0.0.1
- Then open
- http://127.0.0.1:8001/index.html
- Install needed
- No
- Internet needed
- No
- API key needed
- No
- AI at runtime
- No
- Tested on
- Safari (preferred; 62 browser checks passed on guided-v21, 2026-11-19)
- Saves
- none; session progress is in memory and resets on page reload
- Accessibility
- Keyboard pickup/move/drop, visible focus, textual feedback, reduced-motion support and live announcements implemented. Safari DOM/control checks pass on guided-v21. Dedicated screen-reader, narrow-layout, full manual playthrough and download-click acceptance remain pending.
Verification
- Tests
- 53 passing · node tests/run.mjs
- Facts from
- Owner-supplied U13–U15 captures, U13/U14 repair outcomes and U15 replacement guidance in specs/001-noc-ninja-training/evidence/, Source-reviewed questions and evidence limits in data/cases.json
- Known limits
- Three captured incidents, twelve decisions and 42 recorded outputs; no generated telemetry, live device operations or runtime AI.
- Spine optic replacement resolved U13/U14 according to the operator; U15 has no reported repair result. Numerical post-repair measurements and stability intervals were not supplied.
- Historical counters alone do not establish current rates.
- Ten minutes is an unmeasured planning target.
- Dedicated screen-reader, narrow-layout, full manual playthrough and download-click acceptance remain pending. Other browsers have not been requalified for this version.
- License
- Apache-2.0
Who it is for.
A guided patrol through three Arista EOS link incidents, with twelve decisions and 42 supplied command captures. Compare both endpoints, inspect optical readings and fault histories, and explain a repair sequence. Each patrol shuffles incidents, questions and answer positions while keeping the order stable during play.
Drag the shuriken to an answer target, or use Space to pick up, arrow keys to move and Space to drop. Feedback, reasoning hints and retries support investigation. Review all 14 recordings in each case to reveal its reported repair or recommended action, then download the learning handoff.
Two incidents have operator-confirmed repair outcomes; the third recommends replacing the leaf optic first and considering the spine optic only if both the down link and high interlane bias-current spread persist. Numerical post-repair measurements were not supplied. The game preserves those limits and does not generate telemetry.
THE BUILD STORYHow it was made.
Built with GitHub Spec Kit 1.4.2.dev0 and Codex. The exact model, build duration and token usage were not recorded. The game uses only supplied Arista EOS captures; native JavaScript modules validate the incident library, track evidence review and decisions, shuffle each patrol, and generate downloadable text handoffs. Generated ninja, shuriken, Mount Fuji and Himeji Castle artwork is bundled locally.
ADVICEFor the next builder.
- Collect all 14 required command recordings before adding a case. Keep missing measurements and unreported repair outcomes explicit.
- Keep each case’s original captures and reviewed reasoning together. Do not infer rates from historical counters or claim recovery without evidence.
- Host the repository-root entry page over HTTP(S), preserve relative paths and revision query strings, and verify shuriken input and handoff downloads.
Built a game of your own?
Fill in the ARCADE template with your build agent and send me your GitHub URL.
Add your game →