naftiko: 1.0.0-alpha2
info:
  label: Arcade Engine — LLM
  description: 'LLM surface — 1 operations. Lead operation: Language Model Chat. Self-contained Naftiko capability covering one Arcade business surface.'
  tags:
    - Arcade
    - Engine
    - LLM
  created: '2026-05-22'
  modified: '2026-05-22'
binds:
  - namespace: env
    keys:
      ARCADE_API_KEY: ARCADE_API_KEY
capability:
  consumes:
    - type: http
      namespace: engine-llm
      baseUri: https://api.arcade.dev
      description: Arcade Engine — LLM business capability. Self-contained, no shared references.
      authentication:
        type: bearer
        token: '{{env.ARCADE_API_KEY}}'
      resources:
        - name: v1-chat-completions
          path: /v1/chat/completions
          operations:
            - name: llm-chat
              method: POST
              description: Language Model Chat
              outputRawFormat: json
              outputParameters:
                - name: result
                  type: object
                  value: $.
              inputParameters:
                - name: body
                  in: body
                  type: object
                  required: true
                  description: Request body.
  exposes:
    - type: rest
      namespace: engine-llm-rest
      port: 8080
      description: REST adapter for Arcade Engine — LLM. One Spectral-compliant resource per consumed operation, prefixed with /v1.
      resources:
        - path: /v1/chat/completions
          name: v1-chat-completions
          description: REST surface for /v1/chat/completions.
          operations:
            - method: POST
              name: llm-chat
              description: Language Model Chat
              call: engine-llm.llm-chat
              with:
                body: rest.body
              outputParameters:
                - type: object
                  mapping: $.
    - type: mcp
      namespace: engine-llm-mcp
      port: 9090
      transport: http
      description: MCP adapter for Arcade Engine — LLM. One tool per consumed operation, routed inline through this capability's consumes block.
      tools:
        - name: language-model-chat
          description: Language Model Chat
          hints:
            readOnly: false
            destructive: false
            idempotent: false
          call: engine-llm.llm-chat
          with:
            body: tools.body
          outputParameters:
            - type: object
              mapping: $.