ShedBoxAI for Sentiment Analysis

data_sources:
  reviews:
    type: csv
    path: "customer_reviews.csv"

ai_interface:
  model:
    type: rest
    url: "https://api.anthropic.com/v1/messages"
    method: POST
    headers:
      x-api-key: "${ANTHROPIC_API_KEY}"
      Content-Type: "application/json"
    options:
      model: "claude-sonnet-4-20250514"

  prompts:
    sentiment:
      system: "You are a sentiment analysis expert."
      user_template: |
        Analyze the sentiment of these reviews:

        {% for review in reviews %}
        Review {{ loop.index }}: {{ review.text }}
        {% endfor %}

        For each review, provide:
        - Sentiment: positive/negative/neutral
        - Score: 1-10
        - Key themes

output:
  type: file
  path: "sentiment_analysis.json"
  format: json

data_sources:
  mentions:
    type: rest
    url: "https://api.twitter.com/2/tweets/search/recent"
    headers:
      Authorization: "Bearer ${TWITTER_BEARER_TOKEN}"
    options:
      params:
        query: "@yourbrand"
    response_path: "data"

ai_interface:
  model:
    type: rest
    url: "https://api.anthropic.com/v1/messages"
    method: POST
    headers:
      x-api-key: "${ANTHROPIC_API_KEY}"
      Content-Type: "application/json"
    options:
      model: "claude-sonnet-4-20250514"

  prompts:
    social_sentiment:
      system: "You are a social media analyst."
      user_template: |
        Analyze sentiment of these brand mentions:

        {% for mention in mentions %}
        - {{ mention.text }}
        {% endfor %}

        Categorize by: praise, complaint, question, neutral.
        Provide overall brand sentiment score.

output:
  type: file
  path: "social_sentiment.md"
  format: json

data_sources:
  tickets:
    type: csv
    path: "support_tickets.csv"

ai_interface:
  model:
    type: rest
    url: "https://api.anthropic.com/v1/messages"
    method: POST
    headers:
      x-api-key: "${ANTHROPIC_API_KEY}"
      Content-Type: "application/json"
    options:
      model: "claude-sonnet-4-20250514"

  prompts:
    analyze:
      system: "You are a customer experience analyst."
      user_template: |
        Analyze these support tickets:

        {% for ticket in tickets %}
        Ticket #{{ ticket.id }}: {{ ticket.subject }}
        {{ ticket.description }}
        {% endfor %}

        Identify:
        - Common issues (ranked by frequency)
        - Sentiment trends
        - Top improvement areas

output:
  type: file
  path: "support_analysis.md"
  format: json