메뉴
Use for the items catalog CLI, ownership API, reason tags, quarantine/reconcile, or items state and migrations.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use for layout-affecting in-game UI work in this repository: new or changed HUDs, game screens, modal/dialogue sheets, menus, bottom navigation, responsive PC/phone layout, floating layers, clipping/scrolling/overflow, hit targets, or visual proof for a changed game UI surface. Forces a short widget/mode inventory before non-trivial edits so agents reuse existing Neotolis UI APIs. Skip for pure copy tweaks, asset swaps, or runtime automation tasks that do not change layout.
Use ONLY after a source-first search (shared asset library + free CC0/OFL sources) finds no fit, to generate real raster art for AI Studio asset workflows: fake shots, icon/source sheets, sprites, UI art, or source images for later asset tools. Generation is the last resort, not the default. Prefer Codex CLI imagegen via `codex exec` and `scripts/codex_imagegen.sh`; use Antigravity/agy CLI only as fallback. Owns raster creation only; nt-asset-workflow coordinates asset tools, catalog/intake handoff, and ai_studio/quality visual acceptance.
Use when the user asks for a full retrospective or analysis of an AI-assisted work session, chat history, agent run, long task, profiling data, slow progress, repeated failures, weak tool/subagent use, context loss, planning gaps, quality risks, or process improvements. Produces one comprehensive evidence-based session review with bottlenecks, causes, better workflow, and concrete changes for future runs.
Use for layout-affecting in-game UI work in this repository: new or changed HUDs, game screens, modal/dialogue sheets, menus, bottom navigation, responsive PC/phone layout, floating layers, clipping/scrolling/overflow, hit targets, or visual proof for a changed game UI surface. Forces a short widget/mode inventory before non-trivial edits so agents reuse existing Neotolis UI APIs. Skip for pure copy tweaks, asset swaps, or runtime automation tasks that do not change layout.
Use when starting or revising a game concept, first GDD, visual GDD, fake gameplay shot, gameplay loop, design pillars, reference pack, implementation handoff, or design-source cleanup for a specific game. Also use when converting loose ideas into scoped first playable requirements, gameplay data, UI flow, risks, and next implementation steps.
Use when the lead wants to view, test, pick, or play a local AI Studio surface, asset viewer export, static page, or wasm game build from a phone or another device by exposing a self-contained directory through a temporary public cloudflared quick tunnel. Public and unauthenticated while running; never serve secrets or private sources.
| name | nt-game-items |
| description | Use for the items catalog CLI, ownership API, reason tags, quarantine/reconcile, or items state and migrations. |
Workflow router for the split items feature. The invariant core lives in
features/items-core/; content, state, reasons, and seed remain consumer-owned.
features/items-core/README.md for the public surface, tools,
validation, compatibility, and extension boundaries.features/items-core/INSTALL.md for wiring, verification, and removal.features/items-core/feature.json for the versioned machine contract.src/features/items/README.md, then inspect its Lua
manifest/modules, release lock, and state schema as relevant.For single-source Lua catalogs, use the focused semantic CLI. It always takes an explicit game root and delegates evaluation/validation to the canonical sandbox and Snapshot:
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> list
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> inspect --item <item-id> --level-from <n> --level-to <n>
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> dependencies --item <item-id>
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> source --item <item-id>
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> validate
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> validate --affected <item-id>
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> build --out-dir <build-dir>
node ai_studio/dev_environment/python_run.mjs features/items-core/scripts/items_cli.py --project-root <game> seal-receipt
Use chart and requirements for bounded reports; inspect command help for
their filters instead of loading the full Snapshot.
For a supported existing literal, get the hash from source, preview a
level-set, curve-set, override-set, max-level-append, or
max-level-truncate, inspect both diffs, then repeat with --apply. Max-level
edits require explicit generate/columns literals and remain receipt-gated.
Never remove an .items-edit.lock until no writer process is active.
Unsupported source shapes route to direct source/agent editing.
Use batch --patch-file only when every typed operation resolves to the same
Lua file; preview and store its returned inverse batch before --apply.
There is no legacy parser or fallback catalog. Pass the consumer root
explicitly. Follow its README for additions, removals, version bumps,
migrations, and release-lock handling. seal-receipt is the only release
history write path; run it only after the matching migration and checks pass.
nt-game-state-management.items/catalog blob before save load/reconcile or gameplay.features/items-core/README.md.features/items-core/INSTALL.md.features/items-core/feature.json.nt-runtime-automation.nt-quality-checks.