Web Asset Studio
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ASSET_STUDIO_ROOT | No | Where projects are stored | <project dir>/asset-studio |
| ASSET_STUDIO_FFMPEG | No | Used by video_to_frames | ffmpeg on PATH |
| ASSET_STUDIO_DOWNLOADS | No | Folder Chrome saves to, watched by ingest | OS Downloads folder (on Windows, read from the registry, so a relocated Downloads folder works) |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| studio_infoA | Where assets are stored, which downloads folder is watched, ffmpeg availability, and the list of projects. Call first. |
| project_createB | Create an asset project with a style bible that every prompt reuses (this is what keeps a set consistent). |
| project_getC | Project bible, anchor, references, every asset with absolute file paths, and sheets. |
| project_updateC | Change the style bible (merged field by field), pixel settings, or the locked palette. |
| add_referenceA | Copy a reference image (the user's face photo, a style sample) into the project's refs folder. Upload the returned path to ChatGPT. |
| set_anchorA | Mark an approved asset as the anchor: build_prompt then attaches it to every later generation so the character stays the same. |
| build_promptC | Compose the ChatGPT prompt from the bible plus this request, and list the files to attach (anchor first). Paste the prompt verbatim. |
| ingestB | Save a generated image or video into the project (raw/ or video/) and record its prompt. Source: a finished browser download (moved out of Downloads), a URL, or a local path. |
| slice_gridB | Cut a grid image (from task grid/animation) into equal cells; each cell becomes its own asset in frames/. |
| processA | Run steps, in order, on the RAW files of one or more assets and write processed/.png. Always starts from raw, so re-run with new settings freely. Pass every asset of a set in ONE call with the same steps so they come out uniform. Recipes: sticker = remove_bg, trim, outline, fit 512; animation = remove_bg, trim group, pixelate shared. |
| build_spritesheetB | Pack frames into one uniform-cell sprite sheet PNG + JSON atlas (TexturePacker hash, works in Phaser/Pixi; frameWidth/frameHeight for Godot/Unity) + animated GIF preview. |
| video_to_framesC | Extract PNG frames from an ingested video (needs ffmpeg) so they can be processed and packed like other frames. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| asset_studio_workflow | Full procedure for generating consistent asset sets in chatgpt.com with a browser MCP and this server. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 12 tools
Most tools target a distinct stage of the asset pipeline (ingest, process, slice_grid, build_spritesheet), so selection is usually clear. Minor overlap exists between project_get and studio_info (both surface project listings) and between add_reference and ingest (both bring files into a project), but the descriptions clarify the distinction.
All names use snake_case, which helps, but the ordering convention is mixed: noun_verb (project_get, project_create, project_update) alongside verb_noun (add_reference, set_anchor, build_prompt, slice_grid) and bare verbs (process, ingest). It remains readable but is not a predictable single pattern.
Twelve tools is well-scoped for a multi-stage asset pipeline covering project setup, prompt building, ingestion, processing, slicing, and packing. Each tool maps to a distinct operation and none appear redundant.
The surface covers the full lifecycle from project creation and references through ingestion, processing, grid slicing, and sprite sheet export, with clear recipes documented. Gaps are minor, such as no asset/project deletion or rename operations, which agents can likely work around.