agentleFS
HunLuanZhiZhuHunLuanZhiZhu/Game-Studios-Plugin/skills/game-studios-init/SKILL.md

game-studios-init

Skill1 starsChanged 39 days agoMITView on GitHub

Initializes a game project workspace for the Game Studios plugin: scaffolds directories, copies per-project seeds (technical-preferences, architecture registries, settings, rules), generates AGENTS.md with the anti-compression anchor, and merges .gitignore. Invoke when the user asks to initialize or set up a new game project, says 初始化/新项目开工, or right after installing the plugin in a fresh workspace.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Survey first: Check whether AGENTS.md, .studio/, src/, design/, docs/ already exist and whether this is a brownfield project.

    line 34
  2. Very specific:

    Environment inventory (optional functional extension): Run bash <skill-dir>/scripts/env_inventory.sh (add --no-launch to keep it strictly …

    line 51

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Game Studios Project Initialization

Turns the current workspace into a Game Studios project. Idempotent: safe to rerun; existing files are never overwritten — the managed anchor block in AGENTS.md and the managed block in .gitignore are the only content ever replaced in place. All mechanical work is done by a cross-platform (Windows / Linux / macOS) POSIX-sh script (bash + awk — the same runtime the hooks already require), so behavior is identical on every host and every run.

What the script copies (and why)

Workspace pathSourceWhy
.studio/technical-preferences.mdplugin docs/technical-preferences.mdPer-project config with [TO BE CONFIGURED] placeholders; filled by /game-studios:setup-engine; marker file that activates all hooks
.claude/settings.json + .studio/statusline.shskill assetsHost plumbing for Claude Code: permission guardrails + production-stage status line (statusline script lives in the neutral .studio/ dir). ZCode reads no project-level settings file, so none is written for it
.claude/rules/ (11 files)plugin rules/Path-scoped coding standards, auto-enforced by Claude Code (verified). Other hosts: reference-only via the plugin copy (${CLAUDE_PLUGIN_ROOT}/rules/) — no third copy is seeded
.codex/agents/ (51 files)plugin codex/agents/Codex-format subagents (TOML, developer_instructions) generated from agents/*.md by codex/build_agents.sh. Codex plugins cannot bundle subagents, so the workspace copy is the delivery mechanism (loaded only in trusted projects)
CLAUDE.mdskill asset templateStandalone Claude Code entry (collaboration protocol, stack placeholders, @.studio/technical-preferences.md) — not an @AGENTS.md shell. Created if missing; if present, only its managed anchor block is refreshed
docs/registry/architecture.yamlplugin docs/registry/architecture.yamlEmpty scaffold; /game-studios:architecture-decision appends with user approval
docs/architecture/tr-registry.yamlplugin docs/architecture/tr-registry.yamlEmpty scaffold; keeps TR-IDs stable across runs
AGENTS.mdskill asset templateCollaboration protocol, stack placeholders, anti-compression anchor (created only if missing)
directories—src/ assets/ design/ design/gdd/ docs/ docs/architecture/ docs/registry/ tests/ tools/ prototypes/ production/*
.gitignoreskill asset blockFramework ignores appended between managed markers

Procedure

  1. Survey first. Check whether AGENTS.md, .studio/, src/, design/, docs/ already exist and whether this is a brownfield project. If files the script would create already hold user content, confirm scope with the user before applying (copies skip existing files, so nothing is lost).
  2. Locate the script: scripts/init_workspace.sh next to this file. The script resolves the plugin root from its own location; no environment variable is needed. It runs with bash and uses awk only — no sed multiline, no Python.
  3. Run with --check to survey without writing, review the report with the user, then run without flags to apply: bash <skill-dir>/scripts/init_workspace.sh "<workspace-dir>"
  4. Post-run polish (interactive): put the project name/title into AGENTS.md; remind the user that /game-studios:setup-engine populates .studio/technical-preferences.md — until then all hooks stay dormant by design.
  5. Verify: the script prints [created]/[skipped]/[updated] per path and the hooks-marker status. Confirm AGENTS.md and CLAUDE.md contain both <!-- GAME-STUDIOS:BEGIN --> and <!-- GAME-STUDIOS:END -->.
  6. Environment inventory (optional functional extension). Run bash <skill-dir>/scripts/env_inventory.sh (add --no-launch to keep it strictly read-only) and walk the user through the four sections. These checks only enrich the setup — the workspace core (scaffold, rules, registries, hooks marker) is fully functional without any of them; skip freely if the user is not interested.
    • Game engines — detects Godot / Unity / Unreal (PATH, program directories, drive-root scan, registry). If none are found, recommend https://godotengine.org/download (free, open source).
    • Blender / MCP — probes the Blender MCP addon socket on 127.0.0.1:9876 and the blender-mcp bridge CLI. If the socket is down: Blender installed but not running → the script launches Blender and re-probes; if it is still down (or Blender is not installed at all), tell the user that if they need Blender they should follow https://www.blender.org/lab/mcp-server/ (Blender 5.1+, add-on + MCP server).
    • ComfyUI — discovers the install (COMFYUI_HOME → Desktop registry → shallow scan, same order as the comfyui-headless skills), then checks which of the expected Qwen-Image-2.1 and Pixal3D/TRELLIS.2 weight files are actually registered (server API on port 8188 when running, otherwise the registered model roots). image21 needs one file per type; pixal3d needs all 8.
    • Open Design (Windows only) — locates the desktop app (registry InstallLocation) and reports whether its sidecar pipe is live. The plugin's open-design-windows MCP server resolves everything at runtime via mcp/opendesign-mcp.sh (install dir + a live open-design-sidecar-* pipe) — Windows only: auto-discovery uses the Windows registry and Windows named pipes. On macOS/Linux this server is not provided automatically; tell the user to configure the open-design MCP server manually instead (the app can generate its MCP registration) — official site and downloads: https://open-design.ai .
  7. Next steps: /game-studios:setup-engine (engine choice) or the full onboarding flow (game-studios:start).

Notes

  • Brownfield projects: run the script first (it only fills gaps), then game-studios:adopt for format compliance and migration planning.
  • The plugin repo itself is never modified; the workspace owns everything created. Refusing to run inside the plugin repo is built into the script.

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.