Generate a brand-new Codogotchi pet sprite atlas from scratch (from a seed image or a text description): produces the required Codex Tier-1 `spritesheet.webp` and the Lite-Basic Tier-2 `codogotchi-lite-basic-spritesheet.webp` (both 8x9, 192x208px-cell WebP atlases) plus `pet.json`. Use when a user wants to create a new Codogotchi / Maew-style animated desk pet, draw a codogotchi-compatible spritesheet, or hatch a pet when no existing pet art is available.
Generate a brand-new Codogotchi pet sprite atlas from scratch (from a seed image or a text description): produces the required Codex Tier-1 `spritesheet.webp` and the Lite-Basic Tier-2 `codogotchi-lite-basic-spritesheet.webp` (both 8x9, 192x208px-cell WebP atlases) plus `pet.json`. Use when a user wants to create a new Codogotchi / Maew-style animated desk pet, draw a codogotchi-compatible spritesheet, or hatch a pet when no existing pet art is available.
Paths in this skill — scripts/…, references/…, and README.md below are relative to this plugin's root (hatch-codogotchi/, two directories up from this file). cd to the plugin root before running the commands, or prefix each path with it.
I/O policy — scripts own all paths; the skill names none.image_gen cannot reliably store or re-find what it generates, so never tell it to save a row to a named directory and never hunt for an image-generation save folder. Pass whatever local/temp paths the caller picks into the script arguments and chain each script's output forward; the only artifact that must persist is the final magenta atlas the user keys and installs.
No model-side visual inspection. Do not use Computer Use, screenshots, or UI automation to eyeball generated output — the model cannot reliably see its own framing and it burns tokens. Frame geometry is enforced deterministically by slice_grid.py; visual review is the human's, on the script-produced contact sheet and previews.
hatch-codogotchi-codex-and-lite-basic
Generate a brand-new Codogotchi pet from scratch — produces the Codex (Tier 1, required) and Lite-Basic (Tier 2) sheets plus pet.json.
File
Tier
Grid
Dimensions
Rows
spritesheet.webp
1 — Codex
8 × 9
1536 × 1872
9
codogotchi-lite-basic-spritesheet.webp
2 — Lite-Basic
8 × 9
1536 × 1872
9 (incl. ghost)
pet.json
—
—
—
—
Which skill?
Want the full lite set (Basic and Enhanced) for a new pet → SKILL-codex-and-lite-full.md.
Already have a Codex spritesheet.webp → SKILL-lite-basic.md (Basic) / SKILL-lite-enhanced.md / SKILL-soa.md.
Execution model (slot-first): Codex uses its built-in image_gen tool to generate one full animation row per call as 8 invisible equal-size slots, 4 across and 2 down (read left-to-right, top row then bottom), all 8 filled, on the single flat chroma key chosen for this pet (see below). The slots are an invisible placement grid — the prompt says "slots," never "grid/cells," so the model paints one continuous flat field, not 8 separate cards. prepare_pet_run.py also emits a layout-guide.png (4×2 placeholder with slot boundaries, blue safe-margins, centering crosshairs); attach it (and the seed) to every image_gen call as reference-only — it carries placement/centering/safe-margins as a picture, which is what keeps frames from clipping. image_gen need not hit an exact canvas size. Then run slice_grid.py (dimension-tolerant): it slices by fraction, shares one scale across all 8 frames, and emits the exact canonical 1536×208 row strip. Generate one row at a time; never ask for the whole atlas in a single image.
Non-negotiable row gate: finish one row completely (generate grid → slice_grid.py) before generating the next. If a grid comes back clipped, cramped, or off-model, regenerate the whole grid instead of patching forward. Do not compose until every row strip exists. The model does not screenshot or eyeball its own output — slice_grid.py enforces geometry, and prop clarity / face / scale / motion are reviewed by the human on the script-produced contact sheet after composing.
Chroma key — one auto-selected key for the whole pet, keyed by the user later.prepare_pet_run.py picks ONE chroma for the entire pet (every row of every tier shares it): with a seed it auto-selects the canonical key (magenta #FF00FF, blue #0047BB, or green #00B140) farthest from the pet's palette; with no seed (or --chroma magenta) it uses magenta. Reserved accents recolor automatically to avoid the key — the success ✓/stamp (default green) flips to blue under a green key, the ghost (default ethereal blue) flips to warm rose under a blue key — so intentional details are always preserved, never keyed out. This plugin does not key the sheet. After QA, the finished flat-key atlas is handed to the user to key in Codogotchi Studio (https://codogotchi.app/studio), whose tolerance/hue controls separate the art from the key far more reliably than an agent can.
Recommended production pattern: generate the minimum number of distinct keyframes needed for a readable, non-static row, then reuse or mirror earlier stable frames to close the loop when that still looks good in motion. In practice, many rows can be finished faster with ~4 strong keyframes plus a mirrored/reused closure instead of 8 fully independent generations. Treat this as a production shortcut, not a rigid rule.
Read first — the doctrines (full text in README.md + references/animation-rows-lite.md)
0. Motion restraint — row-kind aware (paramount). For standing/status rows, a calm pet with small, smooth motion always beats an expressive one that jitters; when they conflict, choose stability. Anchor the torso, head, hips, and both feet in nearly the same place across all 8 frames; confine motion to one element — the named prop, one arm, or the expression — at low amplitude with short, smooth arcs. For locomotion rows (running-right, running-left), animate a fluid run-in-place stride cycle: keep identity, scale, baseline, and facing stable and the character horizontally centered (no marching across the cell — lateral travel risks clipping), with all 8 frames as distinct, evenly-spaced stride phases that flow. Do not force planted feet, and reject the failure mode where 2–3 frames are nearly identical and then jump. For jumping, animate one full jump cycle across all 8 distinct frames (crouch → push-off → rise → airborne peak → descend → land → settle) with no standing/idle filler frames — give real vertical clearance (feet clearly airborne at the peak, ~24–40 px off baseline, not a shin-high hover that reads as floating), stay horizontally centered, and raise the arms into a gesture that stays below full overhead extension. "No static rows" is a floor (subtle smooth life so frames differ), not a push toward uncontrolled motion.
Prop doctrine — NOT charades. Emotion-mappable states (idle, failed→sad) lead with expression; every other state is carried by one clearly-visible prop — never mimed/"invisible" props, never an A/B prop choice. Same prop, all 8 frames.
Scale consistency. Same character size in all 8 cells of a row. The 8 cells are generated together inside one 4×2 grid, so ask image_gen for a shared head height / body scale across the grid; slice_grid.py shares one scale across the row and the contact sheet surfaces any residual drift for human review.
Visual identity checklist. Every frame must preserve the same age/proportions, hair silhouette, outfit/accessories, palette, and linework as the seed artifact. This is an eyeball pass on the contact sheet — there is no automated identity gate.
Alignment stability. Keep the character on a stable horizontal axis in every 192×208 frame; the pet must not hop left/right. Vertically align ordinary standing rows to a shared bottom baseline near cell_h - 8, not the vertical center. Confirm on the contact sheet / previews.
Plus: don't fake frames by transforming the seed; grid-first, one row at a time; never whole-atlas generation; keep the magenta background perfectly flat.
Quality caveats for the recommended pattern:
Add more unique frames whenever a row's action, emotion, or prop motion reads weakly with mirrored/reused closure.
Never let the speedup hide a prop, flatten the row into near-static motion, or introduce an obvious pop at loop closure.
Passing validation does not automatically make the row acceptable; the human reviews the contact sheet and previews.
Character source
Seed image (recommended): a 192 × 208 neutral standing pose on a solid #FF00FF magenta background.
Text description: generate the Codex idle row first, use its frame 1 as the seed artifact and attach it to every subsequent call as the character anchor.
Row-kind constraints
All rows: flat #FF00FF magenta background with no falloff/shadow/texture/halo · 4×2 grid cells strictly respect 192×208 boundaries · all 8 cells populated, no empty cell · ≥ 8 px padding all sides · one shared character scale across the row · seed is sole style reference. Intentional green props are allowed (keyed later by the user).
Standing/status rows: shared bottom baseline y = 208 − 8 − scaled_h, body and feet anchored, one small moving element, loop closes (frame 8 ≈ frame 1).
Locomotion rows (running-right, running-left): a fluid run-in-place stride cycle — stable scale/baseline/facing, horizontally centered (no lateral travel), all 8 frames distinct and evenly spaced. Do not pin the feet in place; reject teleport jumps and static-static-static-jump timing.
Workflow
# 1. Prepare (seed or --description). Codex + Lite-Basic strip prompt files are written;# each prompt already embeds the prop doctrine, scale rule, and flat #FF00FF background.
python scripts/prepare_pet_run.py --seed <seed> \
--pet-name "My Pet" --style auto --tier codex # then --tier lite-basic# (or --tier all to prep every tier; you generate only codex + lite-basic here)# 2. Use Codex's built-in image_gen tool to generate ONE 4x2 grid per row.# Each grid = 4 cols x 2 rows of 192x208 cells, 8 populated, no empty cell,# flat #FF00FF magenta. image_gen need not hit an exact size. One grid per row;# never ask for the whole atlas at once.# 3. Slice each grid into the canonical 1536x208 row strip (paths caller-chosen/ephemeral)
python scripts/slice_grid.py --input <grid> --out <row-strip>
# slice_grid.py enforces geometry; if a grid is clipped/off-model, regenerate it.# 4. Compose + encode (after ALL rows in a tier) → magenta-background atlas
python scripts/compose_atlas.py --rows-dir <codex-strips> --tier codex --out <work>/spritesheet.png
python scripts/compose_atlas.py --rows-dir <lite-basic-strips> --tier lite-basic --out <atlas-png>
cwebp -lossless -exact <work>/spritesheet.png -o <work>/spritesheet.webp
cwebp -lossless -exact <atlas-png> -o <atlas-webp>
# 5. Slim QA gate for every atlas (runs on the magenta-background sheet)
python scripts/validate_atlas.py --atlas <work>/spritesheet.webp --tier codex --out-json <validation-json>
python scripts/make_contact_sheet.py --atlas <work>/spritesheet.webp --tier codex
python scripts/render_animation_previews.py --atlas <work>/spritesheet.webp --tier codex
python scripts/pre_install_qa_gate.py --atlas <work>/spritesheet.webp --tier codex
python scripts/validate_atlas.py --atlas <atlas-webp> --tier lite-basic --out-json <validation-json>
python scripts/make_contact_sheet.py --atlas <atlas-webp> --tier lite-basic
python scripts/render_animation_previews.py --atlas <atlas-webp> --tier lite-basic
python scripts/pre_install_qa_gate.py --atlas <atlas-webp> --tier lite-basic
# 6. Write pet.json (metadata only; smart defaults for description)
python scripts/prepare_pet_run.py --write-pet-json --run-dir <work>/
pet.json generation — smart defaults for description:--write-pet-json reads run-config.json and writes id, displayName, description, and spritesheetPath. If you passed --description during setup, that text is used as-is. If not, the script auto-generates a smart default from the pet name and style (e.g., "A pixel art My Pet companion for focused work and productivity.") so pet.json is always complete without manual editing.
Final step — keying is the user's, not yours. The composed *.webp atlases still have their flat magenta background. Do not install them. Direct the user to https://codogotchi.app/studio: load each magenta atlas, tune tolerance / edge / spill, export the transparent sheet, and only then install (spritesheet.webp + codogotchi-lite-basic-spritesheet.webp + pet.json) and quit-reopen Codogotchi.
If any frame in a grid is wrong (off pose, scale drift, weak prop, identity drift), regenerate the whole grid for that row, re-run slice_grid.py, and re-run the QA scripts. Fix by regenerating the grid, never by editing individual frames.
Flat #FF00FF background, perfectly uniform across the whole atlas (no falloff/shadow/texture)
Character/content horizontal center is stable for standing/status rows; locomotion rows are a centered in-place stride cycle with stable scale, baseline, facing, and all frames distinct/evenly-spaced (eyeballed)
Stable motion (paramount): standing rows anchor body/feet with one low-amplitude moving element; locomotion rows use a centered in-place stride cycle with no teleport frame
No static rows (gated by validate_atlas.py); each row has subtle distinct motion; loop closes
Each prop-led row shows its single named prop clearly in all 8 frames
Character size consistent across all 8 frames of every row (eyeballed)
Per-frame visual QA passed: same age/proportions, hair silhouette, outfit/accessories, palette, and linework as the seed artifact
Validation JSON, contact sheet, previews, and pre_install_qa_gate.py pass for every atlas
Character consistent across all 18 rows
pet.json present with "id", "displayName", and "spritesheetPath": "spritesheet.webp"
Every row generated as a single 4×2 grid (8 cells, no empty cell), sliced to a 1536×208 strip by slice_grid.py
Before saying done, report: rows generated or regenerated; validation command/result; contact sheet and preview directory paths for every atlas; known compromises. State clearly that the delivered atlases are magenta-background (pre-key) and that the user must key them at https://codogotchi.app/studio before installing. If the tier was completed unusually quickly, state what was compressed, reused, or skipped. Validation alone is not QA.
