What's coming in v1 — scope, highlights, and a thank you

[Donchitos]

Hey everyone,

It's been a while since v0.3.0 and I want to give you a proper look at what's been building in the v1-beta before it merges. This is a big one, and I want you to know what's coming before it lands (and finish testing).

---

What's in v1

Complete story lifecycle

The workflow now has a proper beginning-to-end story pipeline: /story-readiness validates a story before you start, /dev-story drives implementation, and /story-done does an 8-phase completion review including test evidence, design alignment, and surfaces the next ready story automatically.

Full UX/UI pipeline

/ux-design, /ux-review, /team-ui — a dedicated design-to-implementation pipeline for UI and HUD work, with templates for UX specs, HUD layout, accessibility requirements, interaction pattern libraries, and player journey maps.

QA and testing framework

A full suite of QA skills — everything from scaffolding a test framework to detecting flaky tests to triaging your open bug backlog. /qa-plan, /smoke-check, /regression-suite, /test-setup, /test-flakiness, /bug-triage, /team-qa, and more.

Director gates system

All major workflow skills now share named review checkpoints. Gates are defined once and referenced consistently across every phase — no more inline director prompts buried in individual skills.

Gate intensity modes: full / lean / solo

You control how much director oversight you get. full runs all gates, lean skips per-skill reviews but keeps phase gates, solo skips directors entirely and just checks artifact existence. Set it globally during /start, or override per-run with --review [mode] on any skill.

Brownfield adoption

/adopt onboards an existing project into the studio structure. Reads your codebase and generates a project context document, architecture summary, and recommended next steps — no greenfield assumption required.

Traceability registry

Stable requirement IDs that stories and ADRs reference. IDs never get renumbered; the registry is the source of truth across your whole project.

Session resilience

Agent memory, tiered context loading, a post-compaction recovery hook, and a session state checkpoint system that survives context compactions and session crashes.

Godot C# specialist

Dedicated agent for C# code quality in Godot 4, plus language selection in /setup-engine so the right specialists get routed automatically.

Cross-GDD consistency checking

/review-all-gdds runs a holistic pass across all your design documents in parallel. /consistency-check catches contradictions and drift between GDDs before they cause problems downstream.

Session example transcripts

Real end-to-end walkthroughs showing how the skills chain together in practice — brownfield adoption, gate check phase transition, full story lifecycle, UX pipeline, and more. These live in docs/examples/.

Smaller but notable

• Genre-agnostic agents — game-designer, systems-designer, economy-designer, and live-ops-designer no longer assume you're building an RPG
• Worktree isolation for prototyping — throwaway prototype work runs in its own git branch automatically, keeping it safely away from your main codebase
• Directory-scoped instructions — design/, src/, and docs/ each have their own CLAUDE.md so agents get context-appropriate rules depending on where they're working
• The skill system now validates itself — /skill-test is an internal linter and behavioural spec suite for the skills, which is part of what makes v1 feel stable enough to call v1
• Windows toast notifications — when a long-running agent finishes, you get a native desktop notification

---

Scope of v1

v1 is the "complete workflow" release. The goal was a project that a solo developer could take from concept to a shippable build using only the skills in this repo, without hitting a gap where the studio has no answer. That meant closing the story lifecycle, adding real QA tooling, and making director review something you can tune rather than something always in your way.

That said — the project and its workflows are far from finished. There's more depth, more complexity, and more UX improvements to come. This is a milestone, not a destination.

---

On suggestions and feedback

A lot of you have shared ideas, feature requests, and things that felt missing. None of it fell on deaf ears. Most suggestions are either already in the backlog or actively something I want to build. A few things worth addressing directly:

Codex and other agentic platforms — several of you have asked about support beyond Claude Code. This is absolutely intended. It's just a significant amount of work to do properly, so it'll happen over time rather than all at once.

MCP servers and CLI tools — this is on my radar and will likely roll out gradually as more games get worked through the system and I get a clearer picture of where the real pain points are.

"Prompt in, get a game out" — as those of you who've engaged with the system will have noticed, it's not that yet. It's a structured workflow that keeps you in control and guides the process — not an autonomous game generator. That's intentional for now, but I'm aware it's a gap some of you are hoping to close eventually.

Contributing — once v1 is merged and stable, I'm planning to open up PRs and contributions more formally. If you've been sitting on ideas or improvements, that'll be the time.

---

One last thing — show me what you're building

I'd genuinely love to see game examples or prototypes the community has come up with using this system. Even rough ones. Drop them in the replies, link a repo, share a screenshot — whatever you've got. Seeing what people are actually making with this is the best kind of feedback there is.

---

Thank you to everyone who has viewed it, starred it, or interacted with it in any way — it genuinely means a lot. Building in public is only worth it because people actually engage with it, and I appreciate every bit of that.

There's so much to build on from here. This is only the beginning of where this could go, and I hope it helps anyone working on a game — in whatever shape that takes.

[phibu]

When is V1 supposed to come out?

[Shaojie66]

https://github.com/Shaojie66/roll-roll-roll-game

[Shaojie66]

That's great, I'm really looking forward to it😚

[azliwei]

期待 。希望更好的支持更多引擎

[alimgiray-aytar]

Hey man, thanks for the project. It's really good.

[Kazmallar]

Hey @Donchitos !

First off all — thanks for building such an awesome product!

We discovered the project a week ago, and by Monday we were already building our first game from scratch entirely through Claude Code Game Studio. We're on a standard Claude Code subscription with all the usual limits. Stack: Godot 4.3, targeting Android.

We spent about a day and a half writing a detailed GDD:

1. We picked a solid reference project — without one it's really hard to nail down what you actually want from the game.
2. We decided to start with an MVP (roughly 10% of the reference). That means a single level — a tutorial that introduces the core mechanics — with every step wired up to analytics.

The remaining 4 days went into building the prototype. The team basically replicated the core mechanics of the reference project. One thing that really matters: keep your MVP scope ruthlessly tight, otherwise it becomes an endless rabbit hole. We put the reference tutorial under a microscope, documented exactly what we wanted from ours, and went from there. As I'm writing this, the team is already polishing the tutorial mechanics — and everything is covered with unit tests!!!

One curveball: our country has restrictions on Google services, so we had to swap out all the analytics, payments, and ads for local alternatives (Yandex AppMetrica, RuStore SDK, etc.). Not a trivial task, but the team pulled it off!

Next week is all about art — the main goal now is to make the MVP feel as juicy as possible.

I know that's a lot of text, and honestly the story is interesting — but seeing the actual result would be way more fun :)) Fair warning though: no art yet, so right now it's still just a prototype.

emberguard_prototype_040426.mov

[Kazmallar]

One pain point we ran into: keeping track of project status. As the product owner, it's hard to get a clear picture of where development actually stands — which sprint we're on, what milestone we're hitting next. This ends up costing real time just to figure out what's going on.

In a normal workflow, we'd solve this with Jira or some other issue tracker. But how can we handle this here?

From our perspective, this is an important area for the "game dev team" concept to grow into — because we're acting as the client, not the engineers. We don't need to know what's happening inside the black box of development. We just need status. At most, we're making high-level product decisions.

[Gwyllion]

I've been using this repo to make 2 separate games. I've found that you can just tell the team about this problem and it'll work on a solution. I asked it to create a local webpage where project status is tracked visually, and to keep it updated with each commit / each phase transition. I do technical product management for a living, and I found it very refreshing to ask it to create a dashboard with exactly the information I want/need. I wish it was always this easy. Give it a shot! (posting an example of my current polish pass tasklist, there's a lot more on my dashboard)

[Kazmallar]

@Gwyllion amazing, thank you! Genius solutions always somewhere near by ^^

[Gwyllion]

Hey @Donchitos ! Just wanted to give y'all some kudos for this project. I've been using it to clean up an old puzzle game that I never finished due to full responsiveness challenges, as well as for an entirely fresh roguelite math puzzle game. I've been pleasantly surprised about how well the interactions go and how good the results are. The different skills and especially the different specialists that you can engage with in a single session are a huge boon.

I'll post some screenshots here when that game is ready, it's getting there, and it's only 2 weeks in. Hoping for a Google Play closed beta in the next month. No monetization or analytics, just good ole puzzlin'.

Lastly, I don't use Github for my projects, but use Codeberg instead. Not sure if you'd want to have an account over there as well (git is git in the end)