threejs-game-director
Build or upgrade complete Three.js browser games by coordinating gameplay, Mint MCP assets, graphics, UI, debugging, performance, QA, and release.
Works with
--- name: threejs-game-director description: Build or upgrade complete Three.js browser games by coordinating gameplay, Mint MCP assets, graphics, UI, debugging, performance, QA, and release. license: MIT --- # Three.js Game Director Own the end-to-end game outcome. A static scene is not a game, and a working prototype is not automatically polished. ## Route The Work Read the matching sibling `SKILL.md` before that phase: | Need | Skill | | --- | --- | | Game design, loop, controls, camera, physics, feel, scaffold | `../threejs-gameplay-systems/SKILL.md` | | Models, materials, world art, lighting, shaders, VFX, visual quality | `../threejs-visual-systems/SKILL.md` | | HUD, menus, touch controls, responsive interface | `../threejs-game-ui-designer/SKILL.md` | | Picking, manipulation, annotations, configurator-style interaction | `../threejs-interaction-systems/SKILL.md` | | Rendering bugs, loading, input bugs, profiling, optimization | `../threejs-debug-profiler/SKILL.md` | | Browser QA, visual tests, bot playtests, production release | `../threejs-qa-release/SKILL.md` | For broad creation, upgrade, polished, premium, showcase, or release work, use all five. For a narrow fix, use only the relevant specialist plus QA. ## Workflow 1. Inspect the project, current screenshots, controls, runtime, and constraints. 2. For broad gameplay work, define the player promise, primary verb, objective, pressure, reward, fail/retry path, and first level or encounter. 3. Build or repair the playable loop before deep visual polish. 4. If generated assets are needed, follow `../../references/mint-mcp-assets.md`. Mint MCP is the only generated-asset pipeline. Complete finalization, artifact retrieval, project integration, and runtime verification. 5. Upgrade every weak visible surface: hero, threats, rewards, world, materials, lighting, VFX, and UI. Do not hide missing craft with bloom, fog, or glow. 6. Reproduce and measure bugs or performance problems before optimizing. 7. Run player-facing QA and release checks. Continue iterating when the requested quality bar is not met. Use procedural geometry for blockout, collision, repeated support props, debug geometry, deliberate procedural art, or a reported Mint MCP blocker. Never substitute another generation provider. ## Packaged Resources For a new Vite/TypeScript/Three.js game: ```bash python3 <gameplay-skill-dir>/scripts/create_threejs_game.py ./my-game ``` For canvas and renderer evidence: ```bash node <qa-skill-dir>/scripts/inspect-threejs-canvas.mjs --url http://127.0.0.1:5188 ``` ## Completion Gates Broad work requires: - A playable loop proven through real input, objective progress, and fail/retry when the genre has failure. - Build/typecheck and local browser checks with no blocking console/page errors. - Active desktop and mobile screenshots when mobile is in scope. - Nonblank canvas evidence and verification of the changed risky paths. - Readable HUD/menu states, controls, camera, and feedback. - Mint artifact paths and visible runtime integration, or exact Mint MCP blockers, whenever generated assets were needed. Premium, AAA, showcase, polished, complete, release-ready, and "less basic" claims additionally require: - The canonical `../threejs-visual-systems/references/game-visual-scorecard.md` with measured evidence, no category below 2, average at least 2.3, and a fresh-eyes review. - Technical-art budget and renderer diagnostics after graphics changes. - Audio evidence or an explicit blocker for active gameplay. - A visual-test-harness decision; release-ready gameplay also needs a bot playtest or a specific reason it could not run. - Physics-heavy games report engine, timestep, collider strategy, sensors, CCD, and body/collider diagnostics. If a gate fails, continue or report the blocker. Do not soften the quality bar. ## Report Audit For broad or premium work, draft the evidence report and run: ```bash python3 <director-skill-dir>/scripts/audit_reference_report.py --premium /path/to/report.md ``` Add `--physics`, `--audio`, or `--no-design` when applicable. Fix missing evidence before claiming completion. ## Final Response Lead with the outcome. Include changed files, controls, run URL, verification, screenshots/artifacts, Mint links and integrated paths, performance evidence, quality-gate results, and remaining risks. Include the design brief and level/encounter plan only when gameplay scope warrants them.
More Debugging skills
diagnosing-bugs
mattpocock/skills
Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.
explore-code
lllllllama/rigorpilot-skills
Rigor Improve implementation leaf skill for auditable candidate implementation in deep learning research repositories. Use when the researcher explicitly authorizes exploratory work on an isolated branch or worktree to transplant modules, adapt a backbone, add LoRA or adapter layers, replace a head, or stitch together meaningful low-risk migration ideas with rollback-aware records in `explore_outputs/`. Do not use for end-to-end exploration orchestration on top of `current_research`, trusted baseline reproduction, conservative debugging, environment setup, verified contribution claims, or default repository analysis.
safe-debug
lllllllama/rigorpilot-skills
Rigor Debug / Rigor Audit skill for deep learning research work. Use when the user pastes a traceback, terminal error, CUDA OOM, checkpoint load failure, shape mismatch, NaN loss symptom, or training failure and wants conservative diagnosis before any patching, with debug fixes clearly separated from research contributions. Do not use for broad refactoring, speculative adaptation, automatic exploratory patching, or general repository familiarization.

