菜单
Use when designing or repairing a CoGames-style game and the main risk is that the rules, map, roles, or play loop do not yet match the real game or prompt closely enough.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when creating a new mettagrid game from scratch and you need the design-first workflow that writes the rules contract, checks engine fit, and hands a clean build plan to `cg.game.build-game`.
Use when a CoGames-style game already works visually and mechanically and now needs to be split into a composable variant registry, dependency tree, and optional curriculum tree without regressing the default game.
Use when implementing or overhauling a CoGames-style game once the target contract is known, including scaffold, registration, mechanics, assets, and variant factoring.
Use when designing or repairing a CoGames-style game and the main risk is that the rules, map, roles, or play loop do not yet match the real game or prompt closely enough.
Use when a CoGames-style game is mechanically working but still has missing, placeholder, wrongly themed, or non-namespaced art, terrain, or atlas entries.
Use when creating a new mettagrid game from scratch and you need the design-first workflow that writes the rules contract, checks engine fit, and hands a clean build plan to `cg.game.build-game`.
| name | cg.game.core-mechanics |
| description | Use when designing or repairing a CoGames-style game and the main risk is that the rules, map, roles, or play loop do not yet match the real game or prompt closely enough. |
Lock the game loop before art or variant factoring. The target is a mechanically faithful, fully playtested slice that matches the real game, prompt, and expected map/theme in headless and GUI play.
Announce at start: "I’m using cg.game.core-mechanics: I’ll match the real game loop first and prove it with focused playtests before touching art or variant factoring."
Ground the target. Read the current game README/CONTEXT, recipe entrypoints, nearby tests, old Codex logs, and the
real game or prompt. Write down the concrete loop you must preserve: roles, stations, win conditions, map/layout
expectations, and what players should obviously see on screen. For net-new games, this contract should come from
cg.game.new-game; for overhauls, derive it before editing.
Check the existing entrypoint immediately. Use the repo-native play command before editing:
metta play among_us -- render=none max_steps=200
./tools/run.py play overcooked max_steps=40 render=none autostart=true seed=7
metta play mafia render=none max_steps=60 seed=7
Fix the actual mechanics first. Add the nearest focused tests, then implement the smallest playable loop that proves
the game works. Prefer interaction tests near tests/metta/games/<game>/ and recipe-entrypoint checks near
tests/tools/.
Match the real game, not the placeholder. If the map is too large, the roles do not create the right pressure, or the world still looks like the wrong biome (space asteroid instead of kitchen, ship, or village), stop and correct layout/terrain/mechanics before adding more systems.
Run both headless and GUI smoke loops. Headless catches rule regressions fast; GUI catches obvious feel/layout problems. Re-run until the game is readable without explanation.
Ship this stage only when the loop is stable. Do not split variants yet. Do not spend time polishing assets beyond what is necessary to verify the mechanics.
Uses: cf.bop-it, t.run-tests, cg.play, cogame-play -v <variant> (for local variant testing)
Called by: cg.game.build-game, cg.game.new-game
Pairs with: cg.game.generate-assets, cg.game.variant-tree