mediacutlet · April 20, 2025 15:27
diff --git a/HA_FlashBriefing_0.5.yaml b/HA_FlashBriefing_0.5.yaml
 sequence:
  - choose:
      - conditions:
          - condition: template
            value_template: >-
              {{ news_categories_and_sources == 'International News and Global
              Events' }}
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant collecting today's top
                *international* news stories and global events. Search the web
                for the latest headlines **only from reputable sources such as
                CNN, BBC, Reuters, or Associated Press**. Do not include
                U.S.-only stories unless they have significant international
                relevance. Provide 5 headlines and concise summaries (max 265
                characters each). Do not fabricate or summarize beyond what is
                verifiable. If you cannot access these sources, say so. Format
                each item as: 1. [Headline] | [Summary] (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: >-
              {{ news_categories_and_sources == 'United States Politics Top
              Stories' }}
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant collecting today's top *U.S.
                political* news stories. Search the web for verified headlines
                **from sources like CNN, BBC, Reuters, or Associated Press**.
                Focus only on political topics: elections, policy, legislation,
                Supreme Court, and government activity. No summaries beyond
                verifiable content. List 5 items with concise summaries (max 265
                characters). If no political headlines are found from these
                sources, state that. Format as: 1. [Headline] | [Summary]
                (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: >-
              {{ news_categories_and_sources == 'United States Top News
              Excluding Politics' }}
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant collecting today's top
                *non-political* U.S. news stories. Search the web for major
                headlines from CNN, BBC, Reuters, or Associated Press. Exclude
                anything about politics, elections, or legislation. Focus
                instead on events, business, tech, health, or society. Provide 5
                entries with summaries (max 265 characters). Do not fabricate.
                Format as: 1. [Headline] | [Summary] (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: "{{ news_categories_and_sources == 'Weather Related Headlines' }}"
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant gathering today's top
                *weather-related* headlines. Search the web for stories from
                CNN, BBC, Reuters, or Associated Press. Focus on weather events,
                forecasts, natural disasters, and climate-related stories. List
                up to 5 entries, each with a concise summary (max 265
                characters). Do not summarize or add info beyond the source.
                Format as: 1. [Headline] | [Summary] (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: "{{ news_categories_and_sources == 'NASA and Science Headlines' }}"
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant collecting today's top headlines
                related to *NASA, astronomy, space exploration, and scientific
                breakthroughs*. Use only CNN, BBC, Reuters, or Associated Press.
                Avoid speculative stories or fringe science. Provide up to 5
                headlines and short summaries (max 265 characters). Do not make
                anything up. Format as: 1. [Headline] | [Summary] (Source:
                [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: >-
              {{ news_categories_and_sources == 'Media and Entertainment
              Headlines' }}
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                You are a research assistant collecting today's top *media,
                film, TV, and entertainment* news stories. Use only CNN, BBC,
                Reuters, or Associated Press. Avoid celebrity gossip unless it’s
                covered by one of those sources. Provide up to 5 headlines and
                concise summaries (max 265 characters). No fictional content.
                Format as: 1. [Headline] | [Summary] (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
      - conditions:
          - condition: template
            value_template: "{{ news_categories_and_sources == 'Unlisted Topic (Other)' }}"
        sequence:
          - action: conversation.process
            metadata: {}
            data:
              text: >-
                A user has requested a flash briefing on the topic: "{{
                other_topic_name }}". Search the web for current, relevant
                headlines from reputable sources such as CNN, BBC, Reuters, or
                Associated Press. Return up to 5 headlines with short summaries
                (≤265 characters). Only use information found in the results. Do
                not fabricate, speculate, or pull unrelated content. If no
                verifiable stories are found, say so clearly. Format each item
                as: 1. [Headline] | [Summary] (Source: [Name])
              agent_id: conversation.chatgpt
            response_variable: top_news_response
  - variables:
      response:
        international_news: >
          This tool retrieves raw news data that you will use to generate a
          flash briefing.  When the user requests a flash briefing, use this
          tool to fetch recent headlines and short  summaries from reputable
          sources such as CNN, BBC, Reuters, or AP. The user may specify a 
          category, or you may infer the best option based on their request. The
          tool will return up  to 5 headlines relevant to the topic, each paired
          with a short summary and source attribution. Your job is to transform
          this data into a clear, engaging, and factually accurate 
          narrative-style briefing. Do not include any URLs, source names, or
          citations in the final  output. Do not return the stories in a
          numbered or bulleted list. Instead, generate a  conversational,
          paragraph-style summary—like an intelligent and informed family member
          briefly  reviewing today’s top news. Use only the information found in
          the provided results.  Do not invent facts, speculate, editorialize,
          or pull in unrelated content. If the headlines  are sparse or
          unavailable, say so honestly. Prioritize clarity, coherence, and
          truthfulness in  your delivery. When possible, incorporate clear 
          temporal references—such as “today,” “on Wednesday,” “earlier this
          month,”  or “last year”— to help anchor events in time. Only include
          such references  if they are explicitly mentioned or can be reliably
          inferred from the source  summaries.  Here is the news:
          {{top_news_response}}
  - stop: ""
    response_variable: response
 fields:
  news_categories_and_sources:
    selector:
      select:
        options:
          - International News and Global Events
          - United States Politics Top Stories
          - United States Top News Excluding Politics
          - Weather Related Headlines
          - NASA and Science Headlines
          - Media and Entertainment Headlines
          - Unlisted Topic (Other)
    name: News Categories and Sources
    description: >-
      Choose a news source and area of interest from the options below. Doing so
      will return the top five news stories and brief summaries from the
      category and/or source.
    default: Weather Related Headlines
    required: true
  other_topic_name:
    selector:
      text: null
    name: Other Topic Name
    description: >-
      Since an undefined (other) topic was requested for the flash briefing,
      populate this field with the topic, new category, or search intent
      phrasing from the user’s original request. This value will be used to
      guide the downstream news search.
    default: top news in the United States
 alias: "Flash Briefing - Get Latest Top News Headlines "
 description: >-
  This tool retrieves raw news data that you will use to generate a flash
  briefing. When the user requests a flash briefing, use this tool to fetch
  recent headlines and short summaries from reputable news sources (CNN, BBC,
  Reuters, AP). The user may specify a news category; you may choose an option
  which applies best to the user's query. This tool (script) will return up to 5
  headlines relevant to that topic, each paired with a short summary and source
  attribution. Your job is to transform that data into a clear, engaging, and
  factually accurate narrative-style briefing. Do not include any URLs or
  sources in the final output. Only include information found in the results. Do
  not invent details, speculate, or pull in unrelated information. If the
  results are sparse or missing, explain that honestly. Do not invent stories or
  headlines; prioritize truthfulness. Do not return the results in a list
  format, instead deliver a conversational paragraph which spans across the
  topics, as if an intelligent and informed family member was reviewing the top
  news stories of the day. When possible, incorporate clear temporal
  references—such as “today,” “on Wednesday,” “earlier this month,” or “last
  year”— to help anchor events in time. Only include such references if they are
  explicitly mentioned or can be reliably inferred from the source summaries.
	sequence:
	- choose:
	- conditions:
	- condition: template
	value_template: >-
	{{ news_categories_and_sources == 'International News and Global
	Events' }}
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant collecting today's top
	international news stories and global events. Search the web
	for the latest headlines **only from reputable sources such as
	CNN, BBC, Reuters, or Associated Press**. Do not include
	U.S.-only stories unless they have significant international
	relevance. Provide 5 headlines and concise summaries (max 265
	characters each). Do not fabricate or summarize beyond what is
	verifiable. If you cannot access these sources, say so. Format
	each item as: 1. [Headline] \| [Summary] (Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: >-
	{{ news_categories_and_sources == 'United States Politics Top
	Stories' }}
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant collecting today's top *U.S.
	political* news stories. Search the web for verified headlines
	from sources like CNN, BBC, Reuters, or Associated Press.
	Focus only on political topics: elections, policy, legislation,
	Supreme Court, and government activity. No summaries beyond
	verifiable content. List 5 items with concise summaries (max 265
	characters). If no political headlines are found from these
	sources, state that. Format as: 1. [Headline] \| [Summary]
	(Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: >-
	{{ news_categories_and_sources == 'United States Top News
	Excluding Politics' }}
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant collecting today's top
	non-political U.S. news stories. Search the web for major
	headlines from CNN, BBC, Reuters, or Associated Press. Exclude
	anything about politics, elections, or legislation. Focus
	instead on events, business, tech, health, or society. Provide 5
	entries with summaries (max 265 characters). Do not fabricate.
	Format as: 1. [Headline] \| [Summary] (Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: "{{ news_categories_and_sources == 'Weather Related Headlines' }}"
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant gathering today's top
	weather-related headlines. Search the web for stories from
	CNN, BBC, Reuters, or Associated Press. Focus on weather events,
	forecasts, natural disasters, and climate-related stories. List
	up to 5 entries, each with a concise summary (max 265
	characters). Do not summarize or add info beyond the source.
	Format as: 1. [Headline] \| [Summary] (Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: "{{ news_categories_and_sources == 'NASA and Science Headlines' }}"
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant collecting today's top headlines
	related to *NASA, astronomy, space exploration, and scientific
	breakthroughs*. Use only CNN, BBC, Reuters, or Associated Press.
	Avoid speculative stories or fringe science. Provide up to 5
	headlines and short summaries (max 265 characters). Do not make
	anything up. Format as: 1. [Headline] \| [Summary] (Source:
	[Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: >-
	{{ news_categories_and_sources == 'Media and Entertainment
	Headlines' }}
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	You are a research assistant collecting today's top *media,
	film, TV, and entertainment* news stories. Use only CNN, BBC,
	Reuters, or Associated Press. Avoid celebrity gossip unless it’s
	covered by one of those sources. Provide up to 5 headlines and
	concise summaries (max 265 characters). No fictional content.
	Format as: 1. [Headline] \| [Summary] (Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- conditions:
	- condition: template
	value_template: "{{ news_categories_and_sources == 'Unlisted Topic (Other)' }}"
	sequence:
	- action: conversation.process
	metadata: {}
	data:
	text: >-
	A user has requested a flash briefing on the topic: "{{
	other_topic_name }}". Search the web for current, relevant
	headlines from reputable sources such as CNN, BBC, Reuters, or
	Associated Press. Return up to 5 headlines with short summaries
	(≤265 characters). Only use information found in the results. Do
	not fabricate, speculate, or pull unrelated content. If no
	verifiable stories are found, say so clearly. Format each item
	as: 1. [Headline] \| [Summary] (Source: [Name])
	agent_id: conversation.chatgpt
	response_variable: top_news_response
	- variables:
	response:
	international_news: >
	This tool retrieves raw news data that you will use to generate a
	flash briefing. When the user requests a flash briefing, use this
	tool to fetch recent headlines and short summaries from reputable
	sources such as CNN, BBC, Reuters, or AP. The user may specify a
	category, or you may infer the best option based on their request. The
	tool will return up to 5 headlines relevant to the topic, each paired
	with a short summary and source attribution. Your job is to transform
	this data into a clear, engaging, and factually accurate
	narrative-style briefing. Do not include any URLs, source names, or
	citations in the final output. Do not return the stories in a
	numbered or bulleted list. Instead, generate a conversational,
	paragraph-style summary—like an intelligent and informed family member
	briefly reviewing today’s top news. Use only the information found in
	the provided results. Do not invent facts, speculate, editorialize,
	or pull in unrelated content. If the headlines are sparse or
	unavailable, say so honestly. Prioritize clarity, coherence, and
	truthfulness in your delivery. When possible, incorporate clear
	temporal references—such as “today,” “on Wednesday,” “earlier this
	month,” or “last year”— to help anchor events in time. Only include
	such references if they are explicitly mentioned or can be reliably
	inferred from the source summaries. Here is the news:
	{{top_news_response}}
	- stop: ""
	response_variable: response
	fields:
	news_categories_and_sources:
	selector:
	select:
	options:
	- International News and Global Events
	- United States Politics Top Stories
	- United States Top News Excluding Politics
	- Weather Related Headlines
	- NASA and Science Headlines
	- Media and Entertainment Headlines
	- Unlisted Topic (Other)
	name: News Categories and Sources
	description: >-
	Choose a news source and area of interest from the options below. Doing so
	will return the top five news stories and brief summaries from the
	category and/or source.
	default: Weather Related Headlines
	required: true
	other_topic_name:
	selector:
	text: null
	name: Other Topic Name
	description: >-
	Since an undefined (other) topic was requested for the flash briefing,
	populate this field with the topic, new category, or search intent
	phrasing from the user’s original request. This value will be used to
	guide the downstream news search.
	default: top news in the United States
	alias: "Flash Briefing - Get Latest Top News Headlines "
	description: >-
	This tool retrieves raw news data that you will use to generate a flash
	briefing. When the user requests a flash briefing, use this tool to fetch
	recent headlines and short summaries from reputable news sources (CNN, BBC,
	Reuters, AP). The user may specify a news category; you may choose an option
	which applies best to the user's query. This tool (script) will return up to 5
	headlines relevant to that topic, each paired with a short summary and source
	attribution. Your job is to transform that data into a clear, engaging, and
	factually accurate narrative-style briefing. Do not include any URLs or
	sources in the final output. Only include information found in the results. Do
	not invent details, speculate, or pull in unrelated information. If the
	results are sparse or missing, explain that honestly. Do not invent stories or
	headlines; prioritize truthfulness. Do not return the results in a list
	format, instead deliver a conversational paragraph which spans across the
	topics, as if an intelligent and informed family member was reviewing the top
	news stories of the day. When possible, incorporate clear temporal
	references—such as “today,” “on Wednesday,” “earlier this month,” or “last
	year”— to help anchor events in time. Only include such references if they are
	explicitly mentioned or can be reliably inferred from the source summaries.