Slow Ping
Change static routes in working networks to maximize successful ping round-trip time. Three increasingly difficult levels teach weighted paths, independent request/reply routing, longest-prefix matching and TTL limits.

The builder
- Name
- Anthony House
- From
- unknown
- Org
- unknown
- First spec project?
- Yes
- Began at
- NAF AC6
How it was built
- Model
- gpt-6.1-sol+ gpt-6.1-sol
- Harness
- Codex desktop app
- Spec method
- GitHub Spec Kit 1.4.3.dev0
- Spec artifacts
- .specify/memory/constitution.md, StarterKit/CONSTITUTION.md, StarterKit/SPEC.md, StarterKit/PLAN.md, specs/001-maximize-ping-latency/checklists/requirements.md, specs/001-maximize-ping-latency/contracts/interaction.md, specs/001-maximize-ping-latency/data-model.md, specs/001-maximize-ping-latency/plan.md, specs/001-maximize-ping-latency/quickstart.md, specs/001-maximize-ping-latency/research.md, specs/001-maximize-ping-latency/spec.md, specs/001-maximize-ping-latency/tasks.md, specs/002-explicit-path-tunnel/checklists/requirements.md, specs/002-explicit-path-tunnel/contracts/interaction.md, specs/002-explicit-path-tunnel/data-model.md, specs/002-explicit-path-tunnel/plan.md, specs/002-explicit-path-tunnel/quickstart.md, specs/002-explicit-path-tunnel/research.md, specs/002-explicit-path-tunnel/spec.md, specs/002-explicit-path-tunnel/tasks.md, specs/003-static-difficulty/checklists/requirements.md, specs/003-static-difficulty/path-proof.json, specs/003-static-difficulty/plan.md, specs/003-static-difficulty/research.md, specs/003-static-difficulty/spec.md, specs/003-static-difficulty/tasks.md
- Tasks
- 29
- Built
- November 17, 2026
Build environment
- MCP servers
- mcp__cua_repl (integrated-browser computer use; server version unknown)
- Sources & RAG
- User-authored requirements, corrections and supplied NAF_AC6 AGENTS.md development guidelines in this chat
- https://github.com/automateyournetwork/NAF_AC6/blob/main/README.md (local workshop README read)
- https://github.com/automateyournetwork/NAF_AC6/blob/main/WORKSHOP/LAB/README.md (local lab walkthrough read)
- https://github.com/automateyournetwork/NAF_AC6/tree/main/WORKSHOP/LAB/StarterKit (CONSTITUTION.md, SPEC.md, PLAN.md and README.md read locally)
- https://github.com/automateyournetwork/NAF_AC6/blob/main/LICENSE (copied workshop license)
- https://github.com/automateyournetwork/NAF_AC6/blob/main/WORKSHOP/LAB/Examples/PerHopWorks/README.md (worked-example reference only)
- https://github.com/automateyournetwork/NAF_AC6/blob/main/WORKSHOP/LAB/Examples/PerHopWorks/implementation-status.json (example completion status inspected)
- PerHopWorks/tools/browser/package.json and keyboard.mjs (local example developer-tool patterns inspected; these are workshop-relative paths)
- https://github.com/automateyournetwork/NetworkingVideoGameStarterKit (reference and search-result description; full README fetch not recorded)
- https://github.com/github/spec-kit (official CLI installation source, commit f56d514a6bec2d1cbcd3e33dc4387ca6178b4678)
- https://github.github.com/spec-kit/ (documentation search result used)
- https://github.com/github/spec-kit/blob/main/docs/install/uv.md (installation-guide search result and workspace reference)
- https://docs.astral.sh/uv/ (documentation link saved as a reference; full page read not recorded)
- specify version, specify check, specify init --help, gh auth login --help and uv tool install --help (actual local CLI output)
- .specify/templates/constitution-template.md, spec-template.md, plan-template.md, tasks-template.md and checklist-template.md (bundled templates reviewed/resolved)
- .specify/init-options.json and .specify/memory/constitution.md (active project configuration and principles)
- StarterKit/CONSTITUTION.md, StarterKit/SPEC.md and StarterKit/PLAN.md (personalized inputs)
- specs/001-maximize-ping-latency/ (MVP spec including Clarifications, plan, research, data model, UI contract, quickstart, tasks and requirements checklist)
- specs/002-explicit-path-tunnel/ (historical TE experiment spec, plan, research, data model, UI contract, quickstart, tasks and checklist; retired)
- specs/003-static-difficulty/ (active spec, plan, research, tasks, checklist and independent path-proof.json)
- README.md, implementation-status.json, data/levels.json, tests/network.test.js, tests/session.test.js and tools/author-levels.py (active build and independent proof evidence)
- docs/verification.md and docs/tunnel-verification.md (historical verification records)
- docs/static-difficulty-verification.md, docs/domain-results.tap, docs/browser-results.json and docs/arcade-runtime-check.json (current verification records)
- https://www.rfc-editor.org/rfc/rfc1812.html (sections5.2.4 and5.3.1 fetched and searched)
- https://www.rfc-editor.org/rfc/rfc792.html (Echo or Echo Reply fetched and searched)
- https://www.rfc-editor.org/rfc/rfc3209.html (section4.3 strict explicit routes fetched for the retired TE experiment)
- https://automateyournetwork.com/arcade/ARCADE.md (this template, fetched with curl after the web tool could not open it)
- Skills
- speckit-constitution (workspace skill supplied by the user; later installed v1.4.3.dev0 constitution workflow also read)
- speckit-specify (user-supplied workspace skill and installed v1.4.3.dev0 workflow)
- speckit-clarify (installed workflow; prior conversation resolved critical questions)
- speckit-plan (installed workflow; delegated research helper used)
- speckit-tasks (installed workflow)
- speckit-analyze (installed workflow; consistency reviews)
- speckit-implement (installed workflow; task execution and checks)
- loop (SKILL.md inspected to answer an earlier identification question; no loop/heartbeat created or used for the game)
- unified-computer-use (mcp__cua_repl browser interface used for GitHub authentication/repository setup and local preview)
- No Spec Kit extension hooks were configured
- Other tools
- functions.exec orchestration, exec_command shell, apply_patch, write_stdin and view_image
- Native collaboration spawn_agent/followup_task tools (one research helper, reused for protocol review; exact helper model self-reported as gpt-6.1-sol)
- Native web search/open/find tool (official source lookup; not an RFC MCP)
- Git and GitHub browser UI (repository creation, visibility and source control); GitHub CLI OAuth flow was declined because its scopes exceeded the selected repository
- curl (exact Arcade template fetch)
- uv (version unknown) and Specify CLI1.4.3.dev0 (installed from official github/spec-kit commit recorded above)
- Node.js26.0.0 and node:test
- Python3.14.6 (static serving, graph authoring, compile check and manifest processing)
- npm11.12.1 (version shown in installation notice)
- Playwright1.64.1 local Node development package; not Playwright MCP
- Google Chrome157.0.8103.79 (headless developer checks) and Codex integrated browser
- Prettier3.6.2 (developer source formatting)
- PyYAML (version unknown) for frontmatter parsing/validation
- AI assets
- Code-authored native SVG icons, network diagrams and CSS styling; no image, audio or 3D generation tool was used
- Arcade card and verification PNGs are browser screenshots of the game, not image-generator outputs
Runtime: how to play
- Type
- static-web · HTML5, CSS, Browser JavaScript ES modules, Native SVG, Local JSON, Python3.11+ optional local static server
- Start
- python3 start.py
- Then open
- http://127.0.0.1:8001
- Install needed
- No
- Internet needed
- No
- API key needed
- No
- AI at runtime
- No
- Tested on
- Google Chrome157.0.8103.79 (automated browser checks), Codex integrated browser (preview observed; engine version unknown)
- Saves
- Memory only; reload or confirmed session restart clears the session. No localStorage or backend persistence.
- Accessibility
- Keyboard-only campaign using Tab/Enter and native select type-ahead passed; visible focus and text topology/path alternatives checked; reduced-motion behavior and375/760/1024/1440px page-overflow checks passed. Full screen-reader audit, touch-device study and accessibility certification unknown/not performed.
Verification
- Tests
- 20 passing · node --test tests/*.test.js
- Facts from
- RFC1812§5.2.4 (IPv4 route lookup and longest-prefix subset), RFC1812§5.3.1 (TTL decrement and discard), RFC792 Echo/Echo Reply (request and returned reply)
- Known limits
- Deterministic teaching model with fixed symmetric integer one-way delays, zero router processing delay and chosen TTL8
- No ARP, interface subnet negotiation, queueing, congestion, packet loss, dynamic routing, Layer2 switching or live infrastructure
- Finite permitted edits and R3 return-hop restrictions are authored challenge constraints
- Three levels; no arbitrary device placement, wiring, CLI, persistence, accounts, multiplayer or runtime AI
- The MPLS-TE experiment was retired; RFC3209 was used historically but no TE runtime is shipped
- Play duration of the harder revision is unknown; no timed external learner study was recorded
- Full accessibility certification, screen-reader coverage and Safari/Firefox compatibility are unknown/not checked
- No public hosted deployment is recorded; current implementation and manifest commits require a user-authenticated GitHub push
- License
- Apache-2.0
Who it is for.
Slow Ping is a graphical IPv4 routing puzzle for networking learners who want increasingly demanding path-planning challenges. Every topology starts with a working ping; the objective is to maximize its round-trip latency while preserving delivery. Three levels grow from three routers and one editable route to eight routers and sixteen editable entries. Players inspect request and reply paths, discover host-route overrides, and avoid loops and TTL expiry.
THE BUILD STORYHow it was made.
We began with the workshop READMEs and personalized constitution, specification and plan inputs. Spec Kit initialization produced the working constitution and MVP artifacts; clarification, planning, tasks and consistency review preceded implementation. The first static-routing game was followed by a simplified MPLS-TE experiment. The owner then chose static routing with substantially harder levels, so we retired the tunnel runtime and authored six- and eight-router meshes. Python graph search and a separate JavaScript path oracle established maxima of 59, 228 and 349 ms.
The first CLI install hit an authenticated package index; using public PyPI with configuration discovery disabled resolved it. Interactive initialization stalled on an unseen selection and succeeded non-interactively. Headless Chrome dropdown arrows did not work in the test harness, so keyboard checks used native type-ahead. Browser tests drove actual route controls and checked failures, resets, progression, motion and narrow layouts. The current suite has 20 domain/session/content checks and 22 browser checks. A fresh launcher run confirmed a 14 ms baseline and 59 ms edited ping without external requests. The card image is a browser screenshot. GitHub browser authentication did not provide terminal credentials; subsequent commits remain local until pushed.
ADVICEFor the next builder.
- Prove each target independently before implementing its evaluator; include request and reply, prefix precedence and the TTL boundary.
- Increase the routing decisions before adding protocols. Keep draft/application behavior, help and failure explanations visible and testable.
- Record versions, model identifiers, tools and verification as you work. Browser sign-in does not authenticate Git pushes; never put credentials or private machine paths in this manifest.
Built a game of your own?
Fill in the ARCADE template with your build agent and send me your GitHub URL.
Add your game →