SureForge
SkillAn instruction-only skill that makes agents research before they ask, plan before they build, and verify before they call anything done.
- workflow
- verification
- quality
- skill
An agent skill that turns four pixel-art views into a carved voxel volume and renders it as a fake-3D sprite you can spin — with a standalone viewer and PNG export
sprite-stacking is an agent skill for the classic indie-game rendering trick of stacking flat layers to fake depth. The pipeline takes four orthographic views (top, side, front and back, one row), carves a voxel volume that survives only where all four views cover it, cuts the volume into a strip of flat layers ground-first, then renders each layer rotated, squashed for pitch and lifted one pixel. The skill ships a slicer CLI (views_to_slices.py) with flags for layer count (15-20 chunky, 32-48 detailed), view-order matching by proportion or forced reading order, round carving for towers, balloons, pots, side-view facing and shared palette size, plus a standalone viewer.html with a turntable, slice scan, exploded view, drag-to-orbit and 16-direction PNG export; the viewer also accepts a dropped slice strip. If the agent has no image generator it draws the four views itself in a short Python script. Ask for a rotatable pixel-art car, a turntable from a 4-view sheet, or a fake-3D tree for a Godot game. Requires Python 3 with Pillow and NumPy; the skills CLI path needs Node 22.20 or later. The README is explicit that sprite stacking is a long-standing technique — this skill adds the four-view carving and the agent workflow, not the invention. By Daniel White.
Use when a game, toy page or asset pack needs rotatable fake-3D props from a handful of pixel views, or when you want the agent to produce sprite-stack assets end to end without an image model. Skip it for true 3D pipelines or photoreal renders — it is deliberately fake depth.
Via the skills CLI (installs to the current project; -g for every project, -a <agent> to pick one; needs Node 22.20+): npx skills add danielwh2/sprite-stacking Or copy skills/sprite-stacking into your agent's skills folder: Claude Code: ~/.claude/skills/ Codex, Cursor, OpenCode, Gemini CLI, Pi: ~/.agents/skills/ Requirements: python3 with Pillow and NumPy (pip install pillow numpy). Run the slicer directly: python3 skills/sprite-stacking/scripts/views_to_slices.py views.png slices.png --slices 36 --preview check.png --html viewer.html
Rendering rotates and squashes every layer, so a higher --slices count renders proportionally more layers; the README example uses 36. View-order matching runs by proportion by default; use --order reading to force TOP, SIDE, FRONT, BACK. The README credits sprite stacking as prior art and scopes this repo to the four-view carving plus agent workflow. Contributing check: python3 tests/check.py prints ok. MIT. 69 stars as of writing (2026-11-21).
Repository
danielwh2/sprite-stackingCharted
An instruction-only skill that makes agents research before they ask, plan before they build, and verify before they call anything done.
Guizang's agent skill turns your repo's recent updates into a finished promo video — real components, code-rendered animation, an original code-written soundtrack — no video editor involved.
An agent skill that geolocates a photo with nothing readable in it — OSM geometry, elevation skylines and satellite tiles turn “where was this taken?” into a measured answer with an error radius.