tools:
  - name: image_raw_generate
    description: Generate an image using raw AIGC configuration. Usually, check the
      generation status 10 seconds after starting the image generation to allow
      sufficient time for the process.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - prompt
        - aigc_config
      properties:
        prompt:
          type: string
          description: The prompt to generate the image.
        aigc_config:
          type: object
          required:
            - model
          properties:
            ar:
              type: string
              description: Aspect ratio of the image in width:height format, e.g., 1:1, 16:9.
                Must be two positive integers separated by a colon.
            cfg:
              type: number
              description: "Controls how strongly the image follows the prompt. Higher = more
                faithful, but too high may distort. Suggested: 4–10."
            seed:
              type: string
              description: "Controls randomness of image generation. Same seed = same image.
                Use -1 for random. Range: 0-4294967295 or -1."
            step:
              type: number
              description: "Number of sampling steps. Higher steps = clearer, more detailed
                images, but slower. Suggested: 20-40."
            model:
              type: string
              description: Base model for image generation. Can be model name (e.g.,
                animagineXl40Opt.Mt5n), hash (e.g., 6327eca98b), or full title
                (e.g., animagineXl40Opt.Mt5n.safetensors [6327eca98b]).
                Different models affect style and quality.
            width:
              type: number
              description: "Image width (pixels). Set along with width. Range: 64-2000."
            height:
              type: number
              description: "Image height (pixels). Set along with width. Range: 64-2000."
            n_iter:
              type: string
              description: Number of iterations (batch count). Integer (≥1). Defines how many
                times the model runs, generating batch_size images per
                iteration. For example, with n_iter=3 and batch_size=2, a total
                of 6 images are produced.
            sampler:
              enum:
                - DPM++ 2M
                - DPM++ SDE
                - DPM++ 2M SDE
                - DPM++ 2M SDE Heun
                - DPM++ 2S a
                - DPM++ 3M SDE
                - Euler a
                - Euler
                - LMS
                - Heun
                - DPM2
                - DPM2 a
                - DPM fast
                - DPM adaptive
                - Restart
                - DDIM
                - DDIM CFG++
                - PLMS
                - UniPC
                - LCM
              type: string
              description: "Sampling algorithm. Affects image quality, speed, and style.
                Recommended: DPM++ series like DPM++ 2M Karras."
            hr_scale:
              type: string
              description: Upscaling factor for Hires Fix. Float (1.0-4.0, 1 decimal place).
                Determines final image resolution. For example, hr_scale=2.0
                doubles the image size.
            upscaler:
              enum:
                - 4x-UltraSharp
                - R-ESRGAN
                - ESRGAN
              type: string
              description: "Select the upscaling algorithm for increasing image resolution.
                Options: Latent, ESRGAN, R-ESRGAN, 4x-UltraSharp, etc."
            enable_hr:
              enum:
                - "True"
                - "False"
              type: string
              description: "Enable Hires Fix (high-resolution repair). String format: 'True' /
                'False'. Should be 'True' if hr_scale or related parameters are
                set."
            image_url:
              type: string
              description: Upload reference image for ControlNet (e.g., sketch, depth, pose).
                Guides the image generation.
            save_meta:
              enum:
                - "True"
                - "False"
              type: string
              description: "Whether to save metadata (e.g. prompt and settings) into the image
                file. String format: 'True' / 'False'. Useful for
                reproducibility and reference."
            pass_steps:
              type: string
              description: Number of sampling steps in the second pass of Hires Fix. Integer
                format (e.g., 0, 10, 20). Higher values improve detail but
                increase processing time.
            control_image:
              type: string
              description: Upload reference image for ControlNet (e.g., sketch, depth, pose).
                Helps guide image generation.
            control_units:
              type: array
              items:
                type: object
                properties:
                  model:
                    type: string
                    description: Specifies the model used by ControlNet. Accepts model name (e.g.
                      control_v11p_sd15_openpose), hash (e.g. 8e4777ac), or full
                      title (e.g. control_v11p_sd15_openpose [8e4777ac]).
                  module:
                    type: string
                    description: Defines the ControlNet preprocessor module type. String format,
                      e.g., 'ip-adapter_clip_sd15' etc. Determines how the input
                      image is interpreted.
                  weight:
                    type: number
                    description: Defines how strongly the ControlNet input affects the final image.
                      Float (usually 0.0-2.0, 1 decimal place). Higher =
                      stronger control. Default is 1.0.
                  resize_mode:
                    enum:
                      - Just Resize
                      - Crop and Resize
                      - Resize and Fill
                    type: string
                    description: "Defines how the input image is resized to match the generation
                      size. String format, common values: 'Just Resize' - resize
                      directly to target size, 'Crop and Resize' - crop and
                      scale while preserving aspect ratio, 'Resize and Fill' -
                      scale and pad the image to fit"
                  threshold_a:
                    type: number
                    default: 0.5
                    description: First threshold value for preprocessing module (e.g., edge
                      detection). Float format (usually 0.0-255.0, 1 decimal
                      place). Exact behavior depends on the module.
                  threshold_b:
                    type: number
                    default: 0.5
                    description: Second threshold value for preprocessing module (e.g., edge
                      detection). Float format (usually 0.0-255.0, 1 decimal
                      place). Used with threshold_a to define sensitivity.
                  control_mode:
                    type: number
                    default: 0
                    description: "Sets the control priority mode for ControlNet. Integer format: 0 -
                      Balanced (equal influence), 1 - My prompt is more
                      important, 2 - ControlNet is more important"
                  guidance_end:
                    type: number
                    default: 1
                    description: Defines when ControlNet influence stops during sampling. Float
                      (0.0-1.0, 1 decimal place). For example, 0.8 means control
                      ends at 80% of the process.
                  pixel_perfect:
                    enum:
                      - "True"
                      - "False"
                    type: string
                    description: "Enables Pixel Perfect mode, which auto-adjusts preprocessor
                      settings (e.g. resizing). String format: 'True' or
                      'False'. Preserves input image details more accurately."
                  guidance_start:
                    type: number
                    default: 0
                    description: Defines when ControlNet influence starts during sampling. Float
                      (0.0-1.0, 1 decimal place). For example, 0.1 means control
                      starts at 10% of the process.
                additionalProperties: false
              description: An array of ControlNet units. List format, where each item includes
                parameters like input_image, model, module, weight, etc. Allows
                multiple ControlNet modules to be applied together.
            restore_faces:
              enum:
                - "True"
                - "False"
              type: string
              description: "Enable face restoration (typically using GFPGAN or CodeFormer).
                String format: 'True' / 'False'. Improves facial features in
                portraits when enabled."
            denoising_strength:
              type: string
              description: Controls how much the image changes. Float (0.0–1.0, 1 decimal
                place). Higher = more change. 0.3–0.7 is typical, 0.8+ may alter
                too much.
          additionalProperties: false
        negative_prompt:
          type: string
          description: The negative prompt to generate the image.
      additionalProperties: false
  - name: image_collection_generate
    description: Generate an image using a collection's AIGC configuration. You need
      to first check the collection details to understand its variables so you
      can call it correctly. Usually, check the generation status 10 seconds
      after starting the image generation to allow sufficient time for the
      process.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties:
        params:
          type: object
          description: A key-value map used to replace variable placeholders (e.g.
            $gender, $bg) in the collection's prompt. Keys should match variable
            names defined in the collection (without $), and values are their
            replacements.
          additionalProperties:
            type: string
        collection_id:
          type: number
          description: The collection id used to generate the image.
        collection_name:
          type: string
          description: The collection name used to generate the image.
      additionalProperties: false
  - name: image_retry_generate
    description: Retry a failed image generation. Usually, check the generation
      status 10 seconds after starting the image generation to allow sufficient
      time for the process.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - image_id
      properties:
        image_id:
          type: string
          description: The image id to retry.
      additionalProperties: false
  - name: image_state
    description: Get the details of an owned image.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - image_id
      properties:
        image_id:
          type: string
          description: The image id.
      additionalProperties: false
  - name: list_images
    description: Get the list of owned images.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties:
        page:
          type: number
          description: The page number.
        page_size:
          type: number
          description: The page size.
      additionalProperties: false
  - name: generate_task_state
    description: Get the generation state of an image by task id. Usually, check the
      generation status 10 seconds after starting the image generation to allow
      sufficient time for the process.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - task_id
      properties:
        refresh:
          type: string
          default: "true"
          description: When image is generated, refresh the image state.
        task_id:
          type: string
          description: The generation task id.
      additionalProperties: false
  - name: list_loras
    description: Get the list of available loras.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties: {}
      additionalProperties: false
  - name: list_models
    description: Get the list of available models.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties: {}
      additionalProperties: false
  - name: collection_create
    description: Create a new collection.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - name
        - prompt
        - aigc_config
      properties:
        name:
          type: string
          description: The name of the collection.
        prompt:
          type: string
          description: The base prompt string that supports variable placeholders using
            $variable syntax, e.g., 'portrait, 1 $gender, $clothing $style'.
            Variables will be dynamically replaced at image generation time.
        public:
          type: boolean
          default: true
          description: Whether the collection is public.
        variables:
          type: array
          items:
            type: string
          description: A list of variable names (without $) that are used in the prompt
            string, e.g., ['gender', 'clothing', 'style']. These will be
            replaced by values at generation time.
        aigc_config:
          type: object
          required:
            - model
          properties:
            ar:
              type: string
              description: Aspect ratio of the image in width:height format, e.g., 1:1, 16:9.
                Must be two positive integers separated by a colon.
            cfg:
              type: number
              description: "Controls how strongly the image follows the prompt. Higher = more
                faithful, but too high may distort. Suggested: 4–10."
            seed:
              type: string
              description: "Controls randomness of image generation. Same seed = same image.
                Use -1 for random. Range: 0-4294967295 or -1."
            step:
              type: number
              description: "Number of sampling steps. Higher steps = clearer, more detailed
                images, but slower. Suggested: 20-40."
            model:
              type: string
              description: Base model for image generation. Can be model name (e.g.,
                animagineXl40Opt.Mt5n), hash (e.g., 6327eca98b), or full title
                (e.g., animagineXl40Opt.Mt5n.safetensors [6327eca98b]).
                Different models affect style and quality.
            width:
              type: number
              description: "Image width (pixels). Set along with width. Range: 64-2000."
            height:
              type: number
              description: "Image height (pixels). Set along with width. Range: 64-2000."
            n_iter:
              type: string
              description: Number of iterations (batch count). Integer (≥1). Defines how many
                times the model runs, generating batch_size images per
                iteration. For example, with n_iter=3 and batch_size=2, a total
                of 6 images are produced.
            sampler:
              enum:
                - DPM++ 2M
                - DPM++ SDE
                - DPM++ 2M SDE
                - DPM++ 2M SDE Heun
                - DPM++ 2S a
                - DPM++ 3M SDE
                - Euler a
                - Euler
                - LMS
                - Heun
                - DPM2
                - DPM2 a
                - DPM fast
                - DPM adaptive
                - Restart
                - DDIM
                - DDIM CFG++
                - PLMS
                - UniPC
                - LCM
              type: string
              description: "Sampling algorithm. Affects image quality, speed, and style.
                Recommended: DPM++ series like DPM++ 2M Karras."
            hr_scale:
              type: string
              description: Upscaling factor for Hires Fix. Float (1.0-4.0, 1 decimal place).
                Determines final image resolution. For example, hr_scale=2.0
                doubles the image size.
            upscaler:
              enum:
                - 4x-UltraSharp
                - R-ESRGAN
                - ESRGAN
              type: string
              description: "Select the upscaling algorithm for increasing image resolution.
                Options: Latent, ESRGAN, R-ESRGAN, 4x-UltraSharp, etc."
            enable_hr:
              enum:
                - "True"
                - "False"
              type: string
              description: "Enable Hires Fix (high-resolution repair). String format: 'True' /
                'False'. Should be 'True' if hr_scale or related parameters are
                set."
            image_url:
              type: string
              description: Upload reference image for ControlNet (e.g., sketch, depth, pose).
                Guides the image generation.
            save_meta:
              enum:
                - "True"
                - "False"
              type: string
              description: "Whether to save metadata (e.g. prompt and settings) into the image
                file. String format: 'True' / 'False'. Useful for
                reproducibility and reference."
            pass_steps:
              type: string
              description: Number of sampling steps in the second pass of Hires Fix. Integer
                format (e.g., 0, 10, 20). Higher values improve detail but
                increase processing time.
            control_image:
              type: string
              description: Upload reference image for ControlNet (e.g., sketch, depth, pose).
                Helps guide image generation.
            control_units:
              type: array
              items:
                type: object
                properties:
                  model:
                    type: string
                    description: Specifies the model used by ControlNet. Accepts model name (e.g.
                      control_v11p_sd15_openpose), hash (e.g. 8e4777ac), or full
                      title (e.g. control_v11p_sd15_openpose [8e4777ac]).
                  module:
                    type: string
                    description: Defines the ControlNet preprocessor module type. String format,
                      e.g., 'ip-adapter_clip_sd15' etc. Determines how the input
                      image is interpreted.
                  weight:
                    type: number
                    description: Defines how strongly the ControlNet input affects the final image.
                      Float (usually 0.0-2.0, 1 decimal place). Higher =
                      stronger control. Default is 1.0.
                  resize_mode:
                    enum:
                      - Just Resize
                      - Crop and Resize
                      - Resize and Fill
                    type: string
                    description: "Defines how the input image is resized to match the generation
                      size. String format, common values: 'Just Resize' - resize
                      directly to target size, 'Crop and Resize' - crop and
                      scale while preserving aspect ratio, 'Resize and Fill' -
                      scale and pad the image to fit"
                  threshold_a:
                    type: number
                    default: 0.5
                    description: First threshold value for preprocessing module (e.g., edge
                      detection). Float format (usually 0.0-255.0, 1 decimal
                      place). Exact behavior depends on the module.
                  threshold_b:
                    type: number
                    default: 0.5
                    description: Second threshold value for preprocessing module (e.g., edge
                      detection). Float format (usually 0.0-255.0, 1 decimal
                      place). Used with threshold_a to define sensitivity.
                  control_mode:
                    type: number
                    default: 0
                    description: "Sets the control priority mode for ControlNet. Integer format: 0 -
                      Balanced (equal influence), 1 - My prompt is more
                      important, 2 - ControlNet is more important"
                  guidance_end:
                    type: number
                    default: 1
                    description: Defines when ControlNet influence stops during sampling. Float
                      (0.0-1.0, 1 decimal place). For example, 0.8 means control
                      ends at 80% of the process.
                  pixel_perfect:
                    enum:
                      - "True"
                      - "False"
                    type: string
                    description: "Enables Pixel Perfect mode, which auto-adjusts preprocessor
                      settings (e.g. resizing). String format: 'True' or
                      'False'. Preserves input image details more accurately."
                  guidance_start:
                    type: number
                    default: 0
                    description: Defines when ControlNet influence starts during sampling. Float
                      (0.0-1.0, 1 decimal place). For example, 0.1 means control
                      starts at 10% of the process.
                additionalProperties: false
              description: An array of ControlNet units. List format, where each item includes
                parameters like input_image, model, module, weight, etc. Allows
                multiple ControlNet modules to be applied together.
            restore_faces:
              enum:
                - "True"
                - "False"
              type: string
              description: "Enable face restoration (typically using GFPGAN or CodeFormer).
                String format: 'True' / 'False'. Improves facial features in
                portraits when enabled."
            denoising_strength:
              type: string
              description: Controls how much the image changes. Float (0.0–1.0, 1 decimal
                place). Higher = more change. 0.3–0.7 is typical, 0.8+ may alter
                too much.
          additionalProperties: false
        description:
          type: string
          description: The description of the collection.
        train_server:
          type: string
          default: sd
          description: The used aigc server of the collection.
        negative_prompt:
          type: string
          description: The negative prompt.
      additionalProperties: false
  - name: collection_delete
    description: Delete a collection.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties:
        collection_id:
          type: number
          description: The collection id to delete.
        collection_name:
          type: string
          description: The collection name to delete.
      additionalProperties: false
  - name: collection_toggle_public
    description: Toggle the public status of a collection.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties:
        collection_id:
          type: number
          description: The collection id to toggle public.
        collection_name:
          type: string
          description: The collection name to toggle public.
      additionalProperties: false
  - name: collection_list
    description: Get the list of owned collections.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties:
        page:
          type: number
          description: The page.
        page_size:
          type: number
          description: The page size.
        collection_name:
          type: string
          description: The collection name to filter.
      additionalProperties: false
  - name: collection_state
    description: Get the details of an owned collection.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - collection_id
      properties:
        collection_id:
          type: number
          description: The collection id.
      additionalProperties: false
  - name: token_state
    description: Get the x-api-token state.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      properties: {}
      additionalProperties: false
  - name: compress_image
    description: Lossless compression of images (PNG, JPEG, WebP, GIF), support for
      local paths and URLs.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - inputs
      properties:
        inputs:
          type: array
          items:
            type: string
          description: The URLs or full paths to the images to compress.
        quality:
          type: number
          default: 80
          description: The quality of the compressed image.
        outputPath:
          type: string
          description: The full path to save the compressed image.
      additionalProperties: false
  - name: crop_image
    description: Crop images (PNG, JPEG, WebP, GIF) with local path and URL support
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - inputs
      properties:
        top:
          type: number
          default: 0
          description: The left position of the cropped image.
        left:
          type: number
          default: 0
          description: The top position of the cropped image.
        width:
          type: number
          default: 512
          description: The width of the cropped image.
        height:
          type: number
          default: 512
          description: The height of the cropped image.
        inputs:
          type: array
          items:
            type: string
          description: The URLs or full paths to the images to crop.
        outputPath:
          type: string
          description: The full path to save the cropped image.
      additionalProperties: false
  - name: web3_image_tokenization
    description: Tokenize images to become NFT (Non-Fungible Token) on the blockchain.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - image_id
        - recipient
      properties:
        image_id:
          type: number
          description: The image id.
        recipient:
          type: string
          description: The recipient of the NFT, it should be an ethereum address,
            starting at 0x, length 42.
      additionalProperties: false
  - name: web3_job_state
    description: Get the state of a job by job id.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - job_id
      properties:
        job_id:
          type: string
          description: The job id.
        job_type:
          enum:
            - image_tokenization_queue
          type: string
          default: image_tokenization_queue
          description: The type of job.
      additionalProperties: false
  - name: resize_image
    description: Resize images with proportional or fixed dimensions, supports local
      paths and URLs.
    inputSchema:
      type: object
      $schema: http://json-schema.org/draft-07/schema#
      required:
        - inputs
      properties:
        fit:
          enum:
            - cover
            - contain
            - fill
            - inside
            - outside
          type: string
          default: contain
          description: How the image should be resized to fit the target dimensions.
        width:
          type: number
          description: The target width of the resized image in pixels.
        height:
          type: number
          description: The target height of the resized image in pixels.
        inputs:
          type: array
          items:
            type: string
          description: The URLs or full paths to the images to resize.
        outputPath:
          type: string
          description: The full path to save the resized image.
      additionalProperties: false