* Add retrodiffusion
* Detect animation GIF output by magic bytes; add /retrodiffusion to README
The previous animate-mode extension check looked for "gif" in the style
name, but documented animation styles (walk-cycle, idle, attack) never
contain that substring, so GIF responses were always saved as .png and
wouldn't load. Sniff the base64 prefix instead.
Also drop unused `extname` import and add a `/retrodiffusion` row to the
README slash-commands table to match the new user-invocable skill.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: rshtirmer <rshtirmer@gmail.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Strip trailing .,;:!) from captured key values so mid-sentence
pastes like "MESHY_API_KEY=abc123." save correctly
- Anchor all .env grep checks with ^KEY=. to avoid matching comments
or empty assignments
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Skip placeholder values (your-key-here, <key>, etc.) so example
text from skill prompts is never saved as a real secret
- Auto-append .env to the target project's .gitignore when saving keys
- Make additionalContext provider-agnostic (remove hardcoded meshy ref)
- Use `set -a; . .env; set +a` instead of `source .env` so vars are
exported to child processes (node, npx)
- Fix grep/cut command in game-deploy to anchor match and preserve
keys containing = characters
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
API keys pasted in chat previously ended up in Claude's context window.
This adds a UserPromptSubmit hook that detects key patterns, saves them
to .env, and blocks the message so keys never reach the model.
- Add scripts/intercept-api-key.mjs (zero deps, explicit + bare key detection)
- Add hooks.UserPromptSubmit config to settings.json
- Add .env to .gitignore
- Update 6 skill prompts to use KEY=value paste format with .env check-first
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds skill for generating 3D Gaussian Splat environments via World Labs API,
plus worldlabs-generate.mjs helper script. Updates .gitignore to exclude
.herenow/ deployment artifacts.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Meshy AI generates raw GLBs (1-25MB) with uncompressed textures and high
polycounts. New optimize-glb.mjs script automatically reduces file sizes
by 80-95% via texture resize, WebP conversion, and meshopt compression.
- Add scripts/optimize-glb.mjs (standalone, zero-dep, uses npx gltf-transform)
- Auto-optimize after every GLB download in meshy-generate.mjs and find-3d-asset.mjs
- Add MeshoptDecoder to AssetLoader.js template for runtime decompression
- Lower default polycount from 30,000 to 10,000 for game-ready topology
- Add --no-optimize flag to skip optimization when needed
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Fixes 5 issues found during Rock Em Sock Em Robots build:
1. meshy-generate.mjs: rig mode now auto-downloads walking + running
animation GLBs as {slug}-walk.glb and {slug}-run.glb instead of
just printing URLs that expire in 3 days
2. make-game: Meshy API key prompt moved to Step 0 (before scaffolding)
instead of Step 1.5 (after scaffolding is done)
3. make-game: 3D character pipeline is now ONE atomic step
(generate → read meta → rig → auto-download anims) instead of
separate steps requiring manual meta.json parsing between them
4. make-game: integration subagent instructions now mandate orientation
and scale verification — characters must face each other, fit inside
their environment, and be positioned close enough to interact
5. meshyai: rig section documents the 3-file output (rigged GLB +
walk GLB + run GLB) and says "NEVER leave humanoids as static"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New skill and CLI script for generating custom 3D models via Meshy AI API:
- text-to-3d: generate models from text prompts (preview → refine pipeline)
- image-to-3d: generate models from reference images (URL, local file, base64)
- auto-rig: add skeleton to humanoid models for animation
- animate: apply animations to rigged models
- status: check any task's progress
Script handles full lifecycle: submit → poll → download GLB → write meta.json.
Zero npm dependencies, same output format as find-3d-asset.mjs.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add the 3D parallel to the 2D pixel art pipeline:
- scripts/find-3d-asset.mjs: search & download GLBs from Sketchfab, Poly Haven, Poly.pizza
- skills/game-3d-assets: full skill with AssetLoader (SkeletonUtils.clone), OrbitControls
camera, fadeToAction animation crossfade, camera-relative WASD, per-model facingOffset
- skills/add-3d-assets: user-invocable /add-3d-assets command
- 3d-character-library/: 4 animated GLBs (Soldier, Xbot, Robot, Fox) with manifest.json
- templates/threejs-3d: updated with animated character controller, OrbitControls,
AssetLoader, third-person camera follow pattern
- examples/3d-asset-test: working demo with character cycling (C key), preloading,
world props (barrels, crates), animated enemies
- make-game pipeline: Step 1.5 now works for 3D — character selection from library,
world object search via find-3d-asset.mjs, full subagent instructions
Key learnings baked into skill:
- SkeletonUtils.clone() is mandatory for skeletal models (regular .clone breaks → T-pose)
- Model facing varies: facingOffset per character (Soldier -Z needs +PI, Robot +Z needs 0)
- OrbitControls + target-follow is the proven third-person pattern (from official three.js example)
- Preload all GLBs with Promise.all on startup for instant character cycling
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Introduces the Node.js character build pipeline using face-api.js for
face detection and @imgly/background-removal-node for ML background
removal. Includes the trump-mog example game demonstrating South Park
photo-composite characters with Trump and Biden.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add validate-architecture.mjs that statically checks game source files for
required patterns (render_game_to_text, GameState.reset, SAFE_ZONE, button
pattern, EventBus events) with magic number warnings. Separate Phaser and
Three.js variants with appropriate checks for each framework.
Add --restart-cycles N and --restart-key flags to iterate-client.js for
automated multi-restart testing. Polls for game_over state, presses restart,
then verifies score reset and mode transition across N cycles.
Both scripts copied to phaser-2d and threejs-3d templates with npm validate
script added to each package.json.
Co-Authored-By: Claude <noreply@anthropic.com>
The iterate-client.js and verify-runtime.mjs scripts were defined outside
templates and never used in the make-game pipeline. This moves them into
both templates as the single source of truth, fixes the `playwright` →
`@playwright/test` import mismatch, adds robust Playwright prerequisite
checking, and inserts Phase 2.5 (iterate check) into the verification
protocol so every pipeline step gets screenshots + game state feedback.
Also adds render_game_to_text(), advanceTime(), and iterate client docs
to templates, game-qa skill, and CLAUDE.md.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Max Bridgland