Skip to main content
Glama

Scenario.com MCP Server

MCP server providing access to Scenario.com's generative AI API - text-to-image, image-to-image, model training, upscaling, and 70+ other tools.

Installation

Add to your MCP client config:

{
  "mcpServers": {
    "scenario": {
      "command": "npx",
      "args": ["-y", "scenario.com-mcp-server"],
      "env": {
        "SCENARIO_API_KEY": "your_api_key_here",
        "SCENARIO_API_SECRET": "your_api_secret_here"
      }
    }
  }
}

Get your API credentials from https://app.scenario.com/ → Settings → API Keys

Related MCP server: Ideogram MCP Server

Alternative: Session-based Credentials

Don't want to use env vars? Just use:

{
  "mcpServers": {
    "scenario": {
      "command": "npx",
      "args": ["-y", "scenario.com-mcp-server"]
    }
  }
}

Then in your MCP client:

Use set-credentials with API key api_xxx and secret yyy

Available Tools

Setup:

  • set-credentials - Configure API access for session

  • get-credentials-status - Check credential status

Image Generation:

  • post-txt2img-inferences - Generate from text

  • post-img2img-inferences - Generate from image

  • post-controlnet-inferences - ControlNet generation

  • post-inpaint-inferences - Inpainting

  • post-upscale-inferences - Upscale images

  • post-remove-background-inferences - Remove backgrounds

Asset Management:

  • get-assets - List assets

  • delete-asset - Delete assets

  • get-models - List models

Model Training:

  • post-models - Create model

  • put-models-train-by-model-id - Train model

  • post-models-training-images-by-model-id - Add training images

...and 60+ more tools. Use list_tools to see all.

License

ISC

Available Tools

107 tools
copy-asset-by-asset-idC

Duplicate an asset

Inputs: assetId (required), originalAssets, targetProjectId.

delete-assetC

Delete multiple assets

Inputs: assetIds (required).

delete-assets-by-collection-idC

Remove assets from a specific collection

Inputs: collectionId (required), assetIds (required).

delete-collections-by-collection-idC

Delete a collection

Inputs: collectionId (required).

delete-model-preset-by-model-id-and-preset-idC

Delete a preset for the given modelId

Inputs: modelId (required), presetId (required).

delete-models-by-collection-idC

Remove models from a specific collection

Inputs: collectionId (required), modelIds (required).

delete-models-by-model-idC

Delete a model

Inputs: modelId (required).

delete-models-images-by-model-idC

Delete an image

Inputs: modelId (required), ids (required).

delete-models-training-images-by-model-id-and-training-image-idC

Delete the given trainingImageId from the given modelId

Inputs: modelId (required), trainingImageId (required).

delete-workflows-by-workflow-idD

Delete workflow

Inputs: workflowId (required).

get-assetsC

List assets of a project team, or list all public assets

Inputs: updatedBefore, sortDirection, privacy, inferenceId, +15 more.

get-assets-by-asset-idC

Get the details of an asset

Inputs: assetId (required), withEmbedding, originalAssets.

get-canvas-asset-snapshotsC

List snapshots of a canvas type asset

Inputs: assetId (required), pageSize, paginationToken.

get-collectionsC

List collections of a team

Inputs: pageSize, paginationToken.

get-collections-by-collection-idB

Get the details of a collection

Inputs: collectionId (required).

get-credentials-statusA

Check if credentials are configured for this session (does not expose actual values).

Inputs: no named parameters.

get-download-assetsC

Retrieve the status and the url of a batch download assets request

Inputs: jobId (required).

get-job-idC

Get job data by job ID

Inputs: jobId (required).

get-jobsC

List all jobs matching the given filters. A job is a synchronous operation or an asynchronous task such as a training, a generation, etc. It offers a unified view of all operations running on the platform along with their status and results.

Inputs: authorId, hideResults, pageSize, type, +3 more.

get-model-presets-by-model-idC

List all presets for the given modelId

Inputs: modelId (required), originalAssets.

get-modelsC

List all models

Inputs: privacy, originalAssets, pageSize, paginationToken, +4 more.

get-models-by-model-idC

Get the details of the given modelId, including its training status and training progress if available

Inputs: modelId (required), originalAssets.

get-models-description-by-model-idC

Get the description of the given modelId

Inputs: modelId (required), originalAssets.

get-models-examples-by-model-idB

List all examples of the given modelId

Inputs: modelId (required), originalAssets.

get-models-scores-prompt-by-model-idC

Get the prompt scores for the given modelId

Inputs: modelId (required), prompt.

get-models-scores-training-dataset-by-model-idC

Get the training dataset scores for the given modelId

Inputs: modelId (required).

get-public-assetsC

List all public assets

Inputs: updatedBefore, sortDirection, modelId, updatedAfter, +9 more.

get-public-assets-by-asset-idC

Get the details of an asset

Inputs: assetId (required).

get-public-modelsC

List all public models

Inputs: updatedBefore, sortDirection, collectionIds, pageSize, +8 more.

get-public-models-by-model-idC

Get the details of the given modelId

Inputs: modelId (required).

get-public-oscu-pricesB

Get the public Prepaid Compute Units (or OSCU for One Shot Compute Units) price details

Inputs: no named parameters.

get-recommendations-modelsC

List recommended models matching the given filters

Inputs: capabilities, excludeModelIds, limit, nextToken, +5 more.

get-tagsC

List all tags in use for the given projectId

Inputs: pageSize, paginationToken.

get-uploadsB

Get the details of an existing upload

Inputs: uploadId (required).

get-usagesB

Provide usage data for the given filters. Such as consumed compute units, number of assets generated, etc. Maximum time range with custom startDate and endDate is 120 days. Granularity is calculated based on the time range.

Inputs: userId, activityOffset, userIds, type, +4 more.

get-workflowsC

List workflows

Inputs: privacy, pageSize, paginationToken.

get-workflows-by-workflow-idC

Get workflow by ID

Inputs: workflowId (required).

lock-asset-by-asset-idC

Lock a canvas

Inputs: assetId (required), lockExpiresAt (required), originalAssets.

post-assetC

Upload an image or canvas

Inputs: name (required), originalAssets, image, canvas, +4 more.

post-asset-get-bulkC

Get multiple assets by their IDs

Inputs: originalAssets, assetIds.

post-caption-inferencesD

Caption image(s)

Inputs: images (required), dryRun, ensureIPCleared, seed, +6 more.

post-collectionC

Create a new collection

Inputs: name (required).

post-controlnet-img2img-inferencesB

Trigger a new image generation in ControlNet + Img2Img mode. The control image is used to guide the generation; it can be a pose, canny map, or similar. The reference image is used to initialize the generation process.

Inputs: modality (required), modelId (required), prompt (required), originalAssets, +25 more.

post-controlnet-inferencesC

Trigger a new image generation in ControlNet mode. The control image is used to guide the generation; it can be a pose, canny map, or similar.

Inputs: modality (required), modelId (required), prompt (required), originalAssets, +24 more.

post-controlnet-inpaint-inferencesC

Trigger a new image generation in ControlNet + Inpaint mode. The control image is used to guide the generation; it can be a pose, canny map, or similar. The mask indicates the area to inpaint in the reference image.

Inputs: modality (required), modelId (required), prompt (required), originalAssets, +28 more.

post-controlnet-inpaint-ip-adapter-inferencesC

Trigger a new image generation in ControlNet + Inpaint + IpAdapter mode. The control image is used to guide the generation; it can be a pose, canny map, or similar. The mask indicates the area to inpaint in the reference image, and the second reference image is used as an IPAdapter to guide the generation process.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +31 more.

post-controlnet-ip-adapter-inferencesC

Trigger a new image generation in ControlNet + IpAdapter mode. The control image is used to guide the generation; it can be a pose, canny map, or similar. The second reference image is used as an IPAdapter to guide the generation process.

Inputs: modality (required), modelId (required), prompt (required), originalAssets, +31 more.

post-controlnet-texture-inferencesC

Trigger a new seamless texture image generation in ControlNet mode. The control image is used to guide the generation; it can be a pose, canny map, or similar.

Inputs: modality (required), modelId (required), prompt (required), originalAssets, +24 more.

post-describe-style-inferencesC

Describe the style of the given images or models.

Inputs: dryRun, ensureIPCleared, images, seed, +5 more.

post-detect-inferencesC

Advanced precision in image generation by transforming visual data from input images into mode maps.

Inputs: image (required), modality (required), originalAssets, dryRun, +7 more.

post-download-assetsC

Request a link to batch download assets (batch limited to 1000 assets)

Inputs: query (required), options (required).

post-download-modelC

Request a link to download the given modelId

Inputs: modelId (required), modelEpoch.

post-embed-inferencesC

Get embeddings from text

Inputs: text (required), originalAssets, dryRun.

post-generate-customA

Generate with any model (Image, Video, Audio, 3d).

You can retrieve the model inputs from the GET /models/{modelId} endpoint.

Note: This endpoint is not available yet for SD1.5, SDXL, Flux.1 and Flux.1-Kontext based models. For these models, use the POST /generate/{inferenceType} endpoint. Ex: POST /generate/txt2img or POST /generate/prompt-editing

Inputs: modelId (required), body (required), dryRun.

post-generative-fill-inferencesC

Generative fill replace the selected mask area content based on the context. Used to erase objects or characters.

Inputs: image (required), originalAssets, dryRun, targetHeight, +8 more.

post-img2img-inferencesC

Trigger a new image generation in Img2Img mode with one reference image that initializes the generation process.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +19 more.

post-img2img-ip-adapter-inferencesC

Trigger a new image generation in Img2Img + IpAdapter mode. The first image is used to initialize the generation, and the second reference image is used as an IPAdapter.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +28 more.

post-img2img-texture-inferencesC

Trigger a new seamless texture image generation in Img2Img mode with one reference image that initializes the generation.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +21 more.

post-inpaint-inferencesB

Trigger a new image generation in Inpaint mode. The mask indicates the area to inpaint in the reference image.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +22 more.

post-inpaint-ip-adapter-inferencesB

Trigger a new image generation in Inpaint + IpAdapter mode. The mask indicates the area to inpaint in the reference image, and the second reference image is used as an IPAdapter to guide the inpainting.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +29 more.

post-job-action-by-job-idC

Trigger an action on a job: cancel

Inputs: jobId (required), action.

post-model-preset-by-model-idC

Create a new preset for the given modelId

Inputs: modelId (required), inferenceId (required), originalAssets, isDefault.

post-modelsC

Create a new model

Inputs: originalAssets, concepts, collectionIds, classSlug, +4 more.

post-models-copy-by-model-idC

Copy the given modelId to a new model, thumbnail, presets, and all of its training images and pairs if any

Inputs: modelId (required), originalAssets, copyExamples, copyAsTrained.

post-models-get-bulkC

Get multiple models by their modelIds

Inputs: originalAssets, settings, trainingImagesPreview, minimal, +3 more.

post-models-training-images-by-model-idC

Add a new training image to the given modelId

Inputs: modelId (required), originalAssets, data, assetId, +3 more.

post-models-transfer-by-model-idC

Transfer (with a copy or a full ownership change) a model to a new owner, including all of its training images

Inputs: modelId (required), destinationProjectId (required), destinationTeamId.

post-model-training-action-by-model-idC

Trigger an action on a model training: cancel

Inputs: modelId (required), action (required).

post-patch-inferencesD

Patch an asset with an image.

Inputs: image (required), originalAssets, dryRun, patch, +5 more.

post-pixelate-inferencesC

Advanced pixelization of an image.

Inputs: image (required), pixelGridSize (required), removeNoise (required), originalAssets, +4 more.

post-prompt-editing-inferencesC

Edit an image with a prompt.

Inputs: image (required), prompt (required), originalAssets, dryRun, +13 more.

post-prompt-inferencesC

Generate, complete or invent new prompts.

Inputs: mode (required), dryRun, ensureIPCleared, image, +8 more.

post-reframe-inferencesC

Reframe a given image to new sizes. Extra space is filled based on the context.

Inputs: image (required), targetHeight (required), targetWidth (required), originalAssets, +12 more.

post-remove-background-inferencesC

Advanced remove-background of an image.

Inputs: image (required), originalAssets, dryRun, backgroundColor, +1 more.

post-restyle-inferencesC

Trigger a restyle process from one sketch image (or other image) and one or more reference style images.

Inputs: image (required), styleImages (required), originalAssets, dryRun, +9 more.

post-search-assetsC

Search for assets. At least one of the following fields must have a value: query, filter, image, or images.

image, images are mutually exclusive.

Inputs: originalAssets, filter, image, imageSemanticRatio, +9 more.

post-search-modelsC

Search for models. At least one of the following fields must have a value: query, filter, image, or images.

image, and images are mutually exclusive.

Inputs: originalModels, filter, image, imageSemanticRatio, +9 more.

post-segment-inferencesC

Trigger the segmentation of an image. The process will create a new Asset with the segmentation mask as a child.

Inputs: image (required), originalAssets, dryRun, checkpoint, +9 more.

post-skybox-base360-inferencesC

Trigger the generation of a 360 skybox seamless image.

Inputs: prompt (required), originalAssets, dryRun, image, +18 more.

post-skybox-upscale360-inferencesC

Trigger the upscaling of an image matching the 360 skyboxes specific geometry.

Inputs: image (required), originalAssets, dryRun, styleFidelity, +11 more.

post-texture-inferencesA

Trigger the conversion of an image texture to different texture maps:

  • Height map

  • Normal map

  • Smoothness map

  • Metallic map

  • Edge map

  • Ambient Occlusion map

The process will create a new Asset with the above texture maps as children + the original image as an Albedo map.

Inputs: texture (required), originalAssets, dryRun, defaultParameters, +6 more.

post-translate-inferencesC

Translate text from one language to english.

Inputs: prompt (required), dryRun, ensureIPCleared, images, +3 more.

post-txt2img-inferencesC

Trigger a new image generation in Txt2Img mode.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +15 more.

post-txt2img-ip-adapter-inferencesC

Trigger a new image generation in Txt2Img mode with one IpAdapter reference image that guides the generation process.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +22 more.

post-txt2img-texture-inferencesC

Trigger a new seamless texture image generation in Txt2Img mode.

Inputs: modelId (required), prompt (required), originalAssets, dryRun, +14 more.

post-uploadsC

Create a temporary upload URL for a file. Support multipart uploads. Return a list of URLs for each part of the file.

Inputs: fileName, fileSize, kind, civitaiModelUrl, +5 more.

post-uploads-actionC

Trigger an action on upload

Inputs: uploadId (required), action (required).

post-upscale-inferencesA

Trigger the upscaling of an image. You can use styles and presets to quickly get results or craft your very own settings.

Note:This endpoint is deprecated and will be removed in the future. Please leverage POST /generate/custom/{modelId} endpoint instead with model_scenario-upscale-v3 modelId for example. See https://docs.scenario.com/docs/upscale-generation for more details.

Inputs: image (required), originalAssets, dryRun, creativityDecay, +18 more.

post-vectorize-inferencesC

Advanced vectorization of an image.

Inputs: image (required), originalAssets, dryRun, mode, +10 more.

post-workflowsD

Create workflow

Inputs: name (required), description (required).

put-asset-by-asset-idC

Update a canvas asset

Inputs: assetId (required), originalAssets, lockId, lockExpiresAt, +5 more.

put-assets-by-collection-idC

Add assets to a specific collection

Inputs: collectionId (required), assetIds (required).

put-assets-tags-by-asset-idC

Add/delete tags on a specific asset

Inputs: assetId (required), add, strict, delete.

put-collections-by-collection-idC

Update the name of a Collection

Inputs: collectionId (required), name (required).

put-model-preset-by-model-id-and-preset-idD

Modify the given presetId

Inputs: modelId (required), presetId (required), isDefault (required), originalAssets.

put-models-by-collection-idC

Add models to a specific collection

Inputs: collectionId (required), modelIds (required).

put-models-by-model-idD

Update the given modelId

Inputs: modelId (required), originalAssets, negativePromptEmbedding, thumbnail, +8 more.

put-models-description-by-model-idC

Update the markdown description of the given modelId

Inputs: modelId (required), description (required), originalAssets.

put-models-examples-by-model-idC

Add/delete/sort examples of the given modelId

Inputs: modelId (required), assetIds (required), originalAssets.

put-models-tags-by-model-idC

Add/delete tags for the given modelId

Inputs: modelId (required), add, strict, delete.

put-models-train-by-model-idC

Trigger the given modelId training

Inputs: modelId (required), originalAssets, dryRun, parameters.

put-models-training-images-by-model-id-and-training-image-idC

Replace the given trainingImageId for the given modelId

Inputs: modelId (required), trainingImageId (required), originalAssets, data, +4 more.

put-models-training-images-pairs-by-model-idC

Replace all training image pairs for the given modelId

Inputs: modelId (required), body (required).

put-workflow-run-by-workflow-idD

Run a workflow

Inputs: workflowId (required).

put-workflows-by-workflow-idC

Update workflow

Inputs: workflowId (required), name, description.

set-credentialsA

Configure your Scenario.com API credentials for this session. You MUST call this tool first before using any other Scenario tools. Get your credentials from https://app.scenario.com/ (Settings → API Keys).

Inputs: api_key (required), api_secret (required).

unlock-asset-by-asset-idC

Unlock a canvas

Inputs: assetId (required), originalAssets, lockId, forceUnlock.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.7/5.0

    Scored across 107 tools

    Disambiguation3/5

    The tool set has clear distinctions between major resource types like assets, models, collections, and workflows, but there is significant overlap within inference tools (e.g., multiple ControlNet variants, img2img variants) that could confuse agents about which specific mode to use. Descriptions help differentiate, but the sheer number of similar-sounding inference endpoints creates ambiguity.

    Naming Consistency5/5

    Tool names follow a highly consistent verb_noun pattern throughout, using HTTP method prefixes (get, post, put, delete) followed by resource paths with hyphens. This uniform structure makes it predictable and easy to parse, even with the large number of tools.

    Tool Count2/5

    With 107 tools, the count is excessive for a single server, making it overwhelming and difficult for agents to navigate efficiently. While the domain (AI image generation and management) is broad, many tools could be consolidated (e.g., multiple inference variants) or handled via parameters rather than separate endpoints.

    Completeness4/5

    The tool surface provides comprehensive CRUD and lifecycle coverage for core resources (assets, models, collections, workflows) and includes a wide range of AI generation and editing operations. Minor gaps exist, such as limited model training management beyond cancel actions, but overall, the set supports most expected workflows in the domain.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers