ClawHub

PixelLab AI Asset Workflow

Other

PixelLab AI asset workflow and API helper for generating pixel-art images, image conversions, rotations, animations, layered animation sprites, modular outfits/equipment, tilesets, isometric tiles, UI assets, prompt enhancement, and consistent game-art packs. Use when users ask Codex or OpenClaw to plan or run PixelLab API workflows, prepare PixelLab payloads, call PixelLab endpoints, poll background jobs, download generated pixel-art assets, inspect account/list/export/tag/delete API routes, or produce game-engine-ready sprite sheets with stable layers, frame order, pivots, and transparency.

Install

openclaw skills install @uncmatteth/pixellab-ai

PixelLab AI

Use PixelLab as an asset workflow system, not just a prompt endpoint. Route each request to the right endpoint family, prepare a JSON payload, run the helper when the user has configured PIXELLAB_API_KEY, and report the endpoint, result file, and downloaded assets.

What This Skill Provides

  • PixelLab API routing for pixel-art image generation, image-to-pixel-art conversion, cleanup, resizing, background removal, rotation, directional packs, animation, layered/modular animation sprites, tilesets, isometric tiles, map objects, UI assets, character assets, object states, and prompt enhancement.
  • A reusable Python helper that sends authenticated PixelLab /v2 requests, writes JSON results, polls background jobs when needed, downloads discovered asset URLs, and saves base64 image responses as local files.
  • Ready-to-edit payload examples for common PixelLab workflows so agents can start from structured JSON instead of inventing request bodies from scratch.
  • Distilled workflow references from PixelLab docs, official video/tutorial coverage, and community prompting notes, organized so agents only load the relevant reference for the task.
  • Codex and OpenClaw setup guidance that keeps each user responsible for their own PixelLab API key. The package does not include, share, or publish a key.

Use this skill when the user wants a finished game-art asset workflow, not just a nicer prompt: choose the endpoint, prepare the payload, run or explain the API call, poll if needed, save the generated files, and report exactly what happened.

Secret Boundary

  • It is safe and required to name the local env var PIXELLAB_API_KEY; never reveal, print, commit, package, or upload the key value stored in it.
  • Do not ask the user to paste the key into chat. Prefer a local env var or OpenClaw skills.entries.pixellab-ai.apiKey secret config.
  • If the key is missing, stop after preparing payloads and setup instructions. Do not fake generated assets.
  • ClawHub uploads should contain this skill package only. Each installer must provide their own PixelLab key locally.

Workflow

  1. Classify the task: create from scratch, transform an image, build a consistent asset family, rotate views, animate, build maps/tiles/UI, or enhance a weak prompt.
  2. Read references/youtube-workflow-playbook.md for maps, tilesets, side-scrollers, UI, rotations, animation, and style-consistency decisions.
  3. Read references/community-discord-workflows.md when the task involves practical prompting, init images, held items, pose libraries, canvas resizing, tileset export, or animation reuse.
  4. Read references/community-discord-tutorials.md when the user asks for tutorial-channel coverage, Godot/MCP workflows, Pixelorama workflows, top-down RPG direction packs, or a one-by-one source trail.
  5. Read references/sprite-animation-layering.md when the task mentions layered sprites, modular outfits, equipment, animation sprites, frame grids, sprite sheets, pivots/origins, hitboxes, compositing, or game-engine import.
  6. Read references/api-coverage-matrix.md when the user asks for full API coverage, Pro endpoints, account/list/export/tag/delete routes, or a route that is not in the ordinary routing rules.
  7. Read references/endpoint-mapping.md for endpoint selection when the route is not obvious.
  8. Build or edit a payload JSON file outside the skill package, or start from examples/.
  9. For transform examples that contain Base64Image placeholders, replace them with caller-owned PNG/JPEG base64 image objects and matching size metadata before running the helper.
  10. Run scripts/pixellab_client.py with python3, an endpoint path, and a caller-owned output directory.
  11. Use --poll for endpoints that return a background job id or when async behavior is expected.
  12. Return a concise summary: endpoint used, payload path, result JSON path, downloaded files, and any PixelLab error code.

Helper Commands

Create a Pixflux image:

python3 pixellab-ai/scripts/pixellab_client.py post /v2/create-image-pixflux \
  --payload-file pixellab-ai/examples/create-image-pixflux.json \
  --result-file ./pixellab-output/pixflux/result.json \
  --download-dir ./pixellab-output/pixflux

Create a tileset and poll if a background job is returned:

python3 pixellab-ai/scripts/pixellab_client.py post /v2/create-tileset \
  --payload-file pixellab-ai/examples/create-tileset.json \
  --poll \
  --result-file ./pixellab-output/tileset/result.json \
  --download-dir ./pixellab-output/tileset

Poll a known background job:

python3 pixellab-ai/scripts/pixellab_client.py get /v2/background-jobs/JOB_ID \
  --poll \
  --result-file ./pixellab-output/job/result.json \
  --download-dir ./pixellab-output/job

Patch tags or delete managed assets:

python3 pixellab-ai/scripts/pixellab_client.py patch /v2/objects/OBJECT_ID/tags \
  --payload '{"tags":["idle","merchant","south"]}' \
  --result-file ./pixellab-output/object-tags/result.json

python3 pixellab-ai/scripts/pixellab_client.py delete /v2/objects/OBJECT_ID \
  --result-file ./pixellab-output/delete-object/result.json

Routing Rules

  • General text-to-image: /v2/create-image-pixflux, /v2/create-image-pixflux-background, /v2/create-image-pixen, /v2/create-image-bitforge, /v2/generate-image-v2, or /v2/generate-with-style-v2.
  • Existing image cleanup or conversion: /v2/image-to-pixelart, /v2/image-to-pixelart-pro, /v2/remove-background, /v2/resize, /v2/edit-image, /v2/edit-images-v2, /v2/inpaint, or /v2/inpaint-v3.
  • Directional views: /v2/rotate for one view or /v2/generate-8-rotations-v3 for ordinary 8-direction packs; reserve /v2/generate-8-rotations-v2 for Pro workflows.
  • Animation: /v2/animate-with-text-v3, /v2/animate-with-text-v2, /v2/animate-with-text, /v2/animate-with-skeleton, /v2/edit-animation-v2, /v2/interpolation-v2, /v2/transfer-outfit-v2, /v2/create-character-state, /v2/characters/animations, /v2/animate-character, or /v2/objects/{object_id}/animations.
  • Maps, tiles, and UI: /v2/create-tileset, /v2/tilesets, /v2/create-tileset-sidescroller, /v2/tilesets-sidescroller, /v2/create-tiles-pro, /v2/tiles-pro/{tile_id}, /v2/create-isometric-tile, /v2/isometric-tiles, /v2/map-objects, or /v2/generate-ui-v2.
  • Characters and objects: /v2/create-character-v3, /v2/create-character-with-4-directions, /v2/create-character-with-8-directions, /v2/create-character-pro, /v2/create-character-state, /v2/create-1-direction-object, /v2/create-8-direction-object, /v2/characters, /v2/characters/{character_id}, /v2/objects, /v2/objects/{object_id}, /v2/objects/{object_id}/states, /v2/objects/{object_id}/select-frames, or /v2/objects/{object_id}/dismiss-review.
  • Account, jobs, and API metadata: /v2/balance, /v2/background-jobs/{job_id}, or /v2/llms.txt.
  • Prompt enhancement: /v2/enhance-pixen-prompt, /v2/enhance-character-v3-prompt, or /v2/enhance-animation-v3-prompt.

Prompting Rules

  • Reuse a seed when the user wants near-variants; use 0 for random variation when the endpoint supports it.
  • Prefer init/reference images for stronger control over color, pose, style, perspective, or family consistency.
  • For maps, describe what should appear in the middle of the selected area.
  • For weak direction control, combine view/direction parameters with prompt wording such as in profile, facing right, or high top-down.
  • For character prompts, prefer labeled feature blocks such as Body, Clothing, Expression, Pose, Accessories, and Aura over one flat sentence.
  • For character edits, inpaint only the trait that must change, prompt the full target sprite, and reuse the original as a low-strength init image.
  • For animations, start with fewer frames when consistency matters more than length.
  • For animation variants, reuse a finished animation as init frames and keep the seed consistent across frame batches.
  • For layered animation sprites, establish the base motion and frame-grid contract before outfits/equipment; isolate outfit/equipment layers with edit-animation; keep frame count, order, canvas size, origin, and transparency identical across every layer.
  • For style packs, reuse style references and explicit palette language.

References

  • references/youtube-workflow-playbook.md: distilled official YouTube workflow knowledge for assets, tiles, maps, UI, rotation, and animation.
  • references/community-discord-workflows.md: distilled PixelLab Discord helpful-post recipes for prompts, init images, tilesets, canvas resizing, and animation reuse.
  • references/community-discord-tutorials.md: PixelLab Discord #tutorials channel coverage, tutorial feed links, and channel-only workflow tips.
  • references/sprite-animation-layering.md: game-ready sprite animation layering, modular outfits, frame-grid contracts, pivots, compositing, and cleanup workflows.
  • references/api-coverage-matrix.md: current OpenAPI route coverage, helper method support, route status, and known finalization gaps.
  • references/official-youtube-index.md: complete official channel inventory with topic tags and transcript status.
  • references/install-and-secrets.md: Codex, OpenClaw, ClawHub, and API-key setup.
  • references/endpoint-mapping.md: endpoint families, known limits, and error handling.
  • references/prompt-cheatsheet.md: prompt patterns for characters, maps, tilesets, UI, and animation.