MetaGPT/.well-known/skills.yaml

skillapi: "0.1.0"

info:
  title: "Agent Skill Specification"
  version: "1.0"

entities:
  Assistant:
    summary: assistant
    description: assistant
    skills:
      - name: text_to_speech
        description: Generate a voice file from the input text, text-to-speech
        id: text_to_speech.text_to_speech
        x-prerequisite:
          configurations:
            AZURE_TTS_SUBSCRIPTION_KEY:
              type: string
              description: "For more details, check out: [Azure Text-to_Speech](https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=tts)"
            AZURE_TTS_REGION:
              type: string
              description: "For more details, check out: [Azure Text-to_Speech](https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=tts)"
            IFLYTEK_APP_ID:
              type: string
              description: "Application ID is used to access your iFlyTek service API, see: `https://console.xfyun.cn/services/tts`"
            IFLYTEK_API_KEY:
              type: string
              description: "WebAPI argument, see: `https://console.xfyun.cn/services/tts`"
            IFLYTEK_API_SECRET:
              type: string
              description: "WebAPI argument, see: `https://console.xfyun.cn/services/tts`"
          required:
            oneOf:
              - allOf:
                - AZURE_TTS_SUBSCRIPTION_KEY
                - AZURE_TTS_REGION
              - allOf:
                - IFLYTEK_APP_ID
                - IFLYTEK_API_KEY
                - IFLYTEK_API_SECRET
        parameters:
          text:
            description: 'The text used for voice conversion.'
            required: true
            type: string
          lang:
            description: 'The value can contain a language code such as en (English), or a locale such as en-US (English - United States).'
            type: string
            enum:
              - English
              - Chinese
            default: Chinese
          voice:
            description: Name of voice styles
            type: string
            default: zh-CN-XiaomoNeural
          style:
            type: string
            description: Speaking style to express different emotions like cheerfulness, empathy, and calm.
            enum:
              - affectionate
              - angry
              - calm
              - cheerful
              - depressed
              - disgruntled
              - embarrassed
              - envious
              - fearful
              - gentle
              - sad
              - serious
            default: affectionate
          role:
            type: string
            description: With roles, the same voice can act as a different age and gender.
            enum:
              - Girl
              - Boy
              - OlderAdultFemale
              - OlderAdultMale
              - SeniorFemale
              - SeniorMale
              - YoungAdultFemale
              - YoungAdultMale
            default: Girl
        examples:
           - ask: 'A girl says "hello world"'
             answer: 'text_to_speech(text="hello world", role="Girl")'
           - ask: 'A boy affectionate says "hello world"'
             answer: 'text_to_speech(text="hello world", role="Boy", style="affectionate")'
           - ask: 'A boy says "你好"'
             answer: 'text_to_speech(text="你好", role="Boy", lang="Chinese")'
        returns:
          type: string
          format: base64

      - name: text_to_image
        description: Create a drawing based on the text.
        id: text_to_image.text_to_image
        x-prerequisite:
          configurations:
            OPENAI_API_KEY:
              type: string
              description: "OpenAI API key, For more details, checkout: `https://platform.openai.com/account/api-keys`"
            METAGPT_TEXT_TO_IMAGE_MODEL_URL:
              type: string
              description: "Model url."
          required:
            oneOf:
              - OPENAI_API_KEY
              - METAGPT_TEXT_TO_IMAGE_MODEL_URL
        parameters:
          text:
            description: 'The text used for image conversion.'
            type: string
            required: true
          size_type:
            description: size type
            type: string
            default: "512x512"
        examples:
          - ask: 'Draw a girl'
            answer: 'text_to_image(text="Draw a girl", size_type="512x512")'
          - ask: 'Draw an apple'
            answer: 'text_to_image(text="Draw an apple", size_type="512x512")'
        returns:
          type: string
          format: base64

      - name: web_search
        description: Perform Google searches to provide real-time information.
        id: web_search.web_search
        x-prerequisite:
          configurations:
            SEARCH_ENGINE:
              type: string
              description: "Supported values: serpapi/google/serper/ddg"
            SERPER_API_KEY:
              type: string
              description: "SERPER API KEY, For more details, checkout: `https://serper.dev/api-key`"
          required:
            allOf:
              - SEARCH_ENGINE
              - SERPER_API_KEY
        parameters:
          query:
            type: string
            description: 'The search query.'
            required: true
          max_results:
            type: number
            default: 6
            description: 'The number of search results to retrieve.'
        examples:
          - ask: 'Search for information about artificial intelligence'
            answer: 'web_search(query="Search for information about artificial intelligence", max_results=6)'
          - ask: 'Find news articles about climate change'
            answer: 'web_search(query="Find news articles about climate change", max_results=6)'
        returns:
          type: string