app:
  description: "Paste a list of Xiaohongshu / Douyin / TikTok / YouTube links (one per line). Each is read with LinkDigest — transcript, the text in every image, key points with checked quotes — then summarised, and the set is compared. Reads only the links you give it; it does not search."
  icon: "📚"
  icon_background: "#FFEAD5"
  mode: workflow
  name: Social media research agent (LinkDigest)
  use_icon_as_answer_icon: false
kind: app
version: 0.1.5
workflow:
  conversation_variables: []
  environment_variables: []
  features:
    file_upload:
      enabled: false
    opening_statement: ""
    retriever_resource:
      enabled: false
    sensitive_word_avoidance:
      enabled: false
    speech_to_text:
      enabled: false
    suggested_questions: []
    suggested_questions_after_answer:
      enabled: false
    text_to_speech:
      enabled: false
  graph:
    edges:
      - id: start-split
        source: start
        target: split
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: start, targetType: code, isInIteration: false }
      - id: split-each
        source: split
        target: each
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: code, targetType: iteration, isInIteration: false }
      - id: each-read
        source: each
        target: read
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: iteration-start, targetType: tool, isInIteration: true, iteration_id: each }
      - id: read-summarise
        source: read
        target: summarise
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: tool, targetType: llm, isInIteration: true, iteration_id: each }
      - id: each-compare
        source: each
        target: compare
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: iteration, targetType: llm, isInIteration: false }
      - id: compare-end
        source: compare
        target: end
        sourceHandle: source
        targetHandle: target
        type: custom
        data: { sourceType: llm, targetType: end, isInIteration: false }
    nodes:
      - id: start
        type: custom
        position: { x: 40, y: 200 }
        data:
          type: start
          title: Start
          variables:
            - variable: urls
              label: Post links, one per line
              type: paragraph
              required: true
              max_length: 4000
            - variable: question
              label: What do you want to know across these posts?
              type: text-input
              required: false
              max_length: 300
      - id: split
        type: custom
        position: { x: 320, y: 200 }
        data:
          type: code
          title: Split lines
          code_language: python3
          code: |
            def main(urls: str) -> dict:
                links = [l.strip() for l in urls.splitlines() if l.strip()]
                return {"links": links[:20]}
          variables:
            - variable: urls
              value_selector: [start, urls]
          outputs:
            links:
              type: array[string]
      - id: each
        type: custom
        position: { x: 600, y: 160 }
        width: 760
        height: 320
        data:
          type: iteration
          title: Read every link
          iterator_selector: [split, links]
          output_selector: [summarise, text]
          output_type: array[string]
          is_parallel: true
          parallel_nums: 3
          error_handle_mode: continue-on-error
          start_node_id: each-start
      - id: each-start
        type: custom-iteration-start
        position: { x: 24, y: 60 }
        parentId: each
        extent: parent
        data:
          type: iteration-start
          title: ""
          isInIteration: true
      - id: read
        type: custom
        position: { x: 140, y: 60 }
        parentId: each
        extent: parent
        data:
          type: tool
          title: LinkDigest · digest_url
          provider_id: jcaiagent7143-ui/linkdigest/linkdigest
          provider_type: builtin
          provider_name: jcaiagent7143-ui/linkdigest/linkdigest
          tool_name: digest_url
          tool_label: Digest a social link
          tool_configurations:
            format: markdown
            breakdown: false
          tool_parameters:
            url:
              type: mixed
              value: "{{#each.item#}}"
          isInIteration: true
          iteration_id: each
      - id: summarise
        type: custom
        position: { x: 440, y: 60 }
        parentId: each
        extent: parent
        data:
          type: llm
          title: Summarise one post
          model:
            provider: ""
            name: ""
            mode: chat
            completion_params:
              temperature: 0.2
          prompt_template:
            - role: system
              text: |
                You summarise one social media post from the digest below. Use only what the digest says; the digest has a Coverage section saying what was not read — mention it if it matters. Keep the post's language. 120 words at most. Start with the source line "Post: <title> — <author> (<platform>)".
            - role: user
              text: |
                {{#read.text#}}
          context:
            enabled: false
          vision:
            enabled: false
          isInIteration: true
          iteration_id: each
      - id: compare
        type: custom
        position: { x: 1420, y: 200 }
        data:
          type: llm
          title: Compare the set
          model:
            provider: ""
            name: ""
            mode: chat
            completion_params:
              temperature: 0.2
          prompt_template:
            - role: system
              text: |
                You have summaries of several social media posts, each from a digest whose quotes were checked against the post. Answer the question if one was given; otherwise write a short comparison: what they agree on, where they differ, what each adds. Cite posts by their "Post:" line. Say plainly when the summaries do not cover something. Keep the posts' language.
            - role: user
              text: |
                Question: {{#start.question#}}

                Summaries:
                {{#each.output#}}
          context:
            enabled: false
          vision:
            enabled: false
      - id: end
        type: custom
        position: { x: 1700, y: 200 }
        data:
          type: end
          title: End
          outputs:
            - variable: report
              value_selector: [compare, text]
            - variable: summaries
              value_selector: [each, output]