CyVerse Data Science Workflow

naftiko: "1.0.0-alpha1"

info:
  label: "CyVerse Data Science Workflow"
  description: >-
    Unified workflow capability for bioinformatics and data science on the CyVerse
    Discovery Environment. Enables researchers, bioinformaticians, and data scientists
    to manage data in iRODS, discover and launch analysis applications, monitor jobs,
    annotate datasets with metadata, share data with collaborators, and request
    persistent identifiers for publication-ready datasets.
  tags:
    - CyVerse
    - Bioinformatics
    - Data Science
    - iRODS
    - Life Sciences
    - Research Computing
    - Open Science
  created: "2026-05-03"
  modified: "2026-05-03"

binds:
  - namespace: env
    keys:
      TERRAIN_JWT_TOKEN: TERRAIN_JWT_TOKEN

capability:
  consumes:
    - import: terrain
      location: ./shared/terrain.yaml

  exposes:
    - type: rest
      port: 8080
      namespace: cyverse-data-science-api
      description: "Unified REST API for CyVerse Discovery Environment data science workflows."
      resources:
        - path: /v1/filesystem
          name: filesystem
          description: "iRODS data store browsing"
          operations:
            - method: GET
              name: list-directory
              description: "List directory contents in iRODS"
              call: "terrain.list-directory"
              with:
                path: "rest.path"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/filesystem/root
          name: filesystem-root
          description: "User root directory"
          operations:
            - method: GET
              name: get-root-directory
              description: "Get user's root iRODS directory"
              call: "terrain.get-root-directory"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/filesystem/search
          name: filesystem-search
          description: "Search data store"
          operations:
            - method: GET
              name: search-filesystem
              description: "Search iRODS for files and directories"
              call: "terrain.search-filesystem"
              with:
                search: "rest.search"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/filesystem/metadata
          name: metadata
          description: "File metadata (AVUs)"
          operations:
            - method: GET
              name: get-file-metadata
              description: "Get AVU metadata for a path"
              call: "terrain.get-file-metadata"
              with:
                path: "rest.path"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/apps
          name: apps
          description: "Application discovery"
          operations:
            - method: GET
              name: list-apps
              description: "Browse available bioinformatics apps"
              call: "terrain.list-apps"
              with:
                search: "rest.search"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/analyses
          name: analyses
          description: "Analysis job management"
          operations:
            - method: GET
              name: list-analyses
              description: "List user's analysis jobs"
              call: "terrain.list-analyses"
              outputParameters:
                - type: object
                  mapping: "$."
            - method: POST
              name: submit-analysis
              description: "Submit a new analysis job"
              call: "terrain.submit-analysis"
              with:
                name: "rest.name"
                app_id: "rest.app_id"
                system_id: "rest.system_id"
                config: "rest.config"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/notifications
          name: notifications
          description: "User notifications"
          operations:
            - method: GET
              name: list-notifications
              description: "List user notifications"
              call: "terrain.list-notifications"
              outputParameters:
                - type: object
                  mapping: "$."

    - type: mcp
      port: 9090
      namespace: cyverse-data-science-mcp
      transport: http
      description: "MCP server for AI-assisted CyVerse Discovery Environment data science workflows."
      tools:
        - name: list-directory
          description: "Browse an iRODS directory in the CyVerse data store"
          hints:
            readOnly: true
            openWorld: true
          call: "terrain.list-directory"
          with:
            path: "tools.path"
          outputParameters:
            - type: object
              mapping: "$."
        - name: get-root-directory
          description: "Get the root directory structure for the CyVerse user account"
          hints:
            readOnly: true
            openWorld: false
          call: "terrain.get-root-directory"
          outputParameters:
            - type: object
              mapping: "$."
        - name: get-file-stats
          description: "Get size, permissions, and modification dates for an iRODS file or directory"
          hints:
            readOnly: true
            openWorld: false
          call: "terrain.get-file-stats"
          with:
            path: "tools.path"
          outputParameters:
            - type: object
              mapping: "$."
        - name: search-filesystem
          description: "Search the CyVerse iRODS data store for files and folders by name"
          hints:
            readOnly: true
            openWorld: true
          call: "terrain.search-filesystem"
          with:
            search: "tools.search"
            type: "tools.type"
          outputParameters:
            - type: object
              mapping: "$."
        - name: move-files
          description: "Move files or directories to a new location in iRODS"
          hints:
            readOnly: false
            destructive: false
            idempotent: false
          call: "terrain.move-files"
          with:
            sources: "tools.sources"
            dest: "tools.dest"
          outputParameters:
            - type: object
              mapping: "$."
        - name: delete-files
          description: "Move files or directories to the user's iRODS trash"
          hints:
            readOnly: false
            destructive: true
            idempotent: false
          call: "terrain.delete-files"
          with:
            paths: "tools.paths"
          outputParameters:
            - type: object
              mapping: "$."
        - name: get-file-metadata
          description: "Get iRODS AVU metadata triples for a file or directory"
          hints:
            readOnly: true
            openWorld: false
          call: "terrain.get-file-metadata"
          with:
            path: "tools.path"
          outputParameters:
            - type: object
              mapping: "$."
        - name: share-data
          description: "Share iRODS data (files/folders) with other CyVerse users"
          hints:
            readOnly: false
            destructive: false
          call: "terrain.share-data"
          with:
            sharing: "tools.sharing"
          outputParameters:
            - type: object
              mapping: "$."
        - name: list-apps
          description: "Browse bioinformatics and data science apps in the Discovery Environment"
          hints:
            readOnly: true
            openWorld: true
          call: "terrain.list-apps"
          with:
            search: "tools.search"
          outputParameters:
            - type: object
              mapping: "$."
        - name: list-analyses
          description: "List the user's analysis job history in the Discovery Environment"
          hints:
            readOnly: true
            openWorld: true
          call: "terrain.list-analyses"
          outputParameters:
            - type: object
              mapping: "$."
        - name: submit-analysis
          description: "Submit a new bioinformatics analysis job using a Discovery Environment application"
          hints:
            readOnly: false
            destructive: false
          call: "terrain.submit-analysis"
          with:
            name: "tools.name"
            app_id: "tools.app_id"
            system_id: "tools.system_id"
            config: "tools.config"
          outputParameters:
            - type: object
              mapping: "$."
        - name: list-notifications
          description: "Get recent notifications about completed analyses and system events"
          hints:
            readOnly: true
            openWorld: true
          call: "terrain.list-notifications"
          outputParameters:
            - type: object
              mapping: "$."