agentleFS
rodrigo-guerraarodrigo-guerraa/roblox-hub/skills/vendored/roblox-to-issues/SKILL.md

roblox-to-issues

Skill2 starsChanged 31 days agoMITView on GitHub

Break a Roblox game plan, PRD or feature idea into small vertical "tracer bullet" slices, each cutting through client, server, remotes, data and UI for one player-facing capability, and each done only when verified in Studio through the Roblox Studio MCP. Use when turning a concept into buildable tickets, splitting work, planning what to build next, or stopping an agent from building a whole game in one pass.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Prefer many thin slices over a few thick ones.

    line 27

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Roblox to Issues (Vertical Slices)

Where this conflicts with the roblox-lessons skill, roblox-lessons wins.

Break a plan into small issues that are each buildable and verifiable on their own. This is the antidote to "build the whole game in one go, everything half-broken."

Where slices live

Use whatever tracker the project already uses: GitHub issues via gh (only after the user approves publishing), or the project's existing plan/task file in the repo. If there is none, ask the user where slices should go; do not invent new tracker folders or config files.

What a vertical slice is

A tracer bullet: the thinnest end-to-end path a player can feel, cutting through every layer it touches (client input, server authority, remotes, saved data, UI), not one horizontal layer.

  • Right (vertical): "Player taps the ore, server grants 1 coin, HUD shows the new balance, balance survives rejoin."
  • Wrong (horizontal): "Build the entire economy module." / "Build all the UI." Nothing is playable until everything is done.

Slice rules

  • Each slice delivers one narrow but complete player-facing capability.
  • Each slice is verifiable in Studio through the MCP on its own; if it can't be, split it.
  • Prefer many thin slices over a few thick ones.
  • Valuable state (currency, inventory, purchases, rewards, progression) is server-authoritative from the first slice, never a client-trust shortcut to fix later.
  • Slice 1 is the tracer bullet: the smallest thing proving the core loop end to end (one action, one reward, one visible next goal).

Slice types

  • AFK: can be built and verified without a human decision. Prefer these.
  • HITL: needs the human: a design approval, a balance or price call, publishing, anything spending Robux, or a test only they can run (real purchases, phones, multi-client sessions). A UI slice with no agreed look is HITL: add a design-approval slice first and block the UI slice on it.

Process

1. Gather context

Work from the plan or conversation. If the user passes a reference (issue number, file path), read it fully. Use the game's own vocabulary.

2. Draft the slices

Core-loop tracer bullet first, then layer outward (persistence, progression, social, monetization, polish). Each later slice builds on one that is already playable.

3. Quiz the user

Present a numbered list. For each slice show: Title (player-facing), Type (AFK/HITL), Blocked by, Verification (the exact MCP steps that prove it), Server-authoritative? (which state it protects). Ask whether granularity, dependencies and types are right. Iterate until approved.

4. Publish

Publish approved slices blockers-first so later slices can reference real IDs. Do NOT close or modify a parent PRD/issue.

<slice-template> ## What to build

The end-to-end behavior: what the player does and what the server does. Not a layer-by-layer implementation.

Layers touched

  • Client: ...
  • Server (authoritative): ...
  • Remotes: ...
  • Data / persistence: ...
  • UI: ...

Done when (verified in Studio via MCP)

  • Code is in the repo and Rojo-synced; a fresh Play (start_stop_play) shows no new errors or warnings in get_console_output.
  • <exact action> produces <exact result>, observed through execute_luau (instances, attributes, captured remote payloads), not assumed.
  • Client-to-server remotes accept a valid payload and silently reject forged ones.
  • Server-to-client syncs fire, including on join.
  • Valuable state is server-validated and survives rejoin (if applicable).
  • UI and world changes checked with screen_capture (UI also at phone size).
  • Temporary test scripts removed. Anything untestable here is named with a short human test.

Blocked by

The blocking slice, or "None - can start immediately". </slice-template>

Status

A slice is done only when every "Done when" box is ticked from real MCP observations. "It compiles" or "the code looks right" is not done. Mark slices needing human approval as built-and-verified until the human signs off. The human pushes commits.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.