Pular para o conteúdo
TakoAPI
Voltar às skills

social-sentiment

por atyachin

DevOps & Cloud
2.8kDownloads informados pelo clawskills.sh
4Estrelas
0Visualizações
0 curtidaEntre para curtir esta skill
—Ainda sem avaliações
Sua avaliação

Resumo

Sentiment analysis for brands and products across Twitter, Reddit, and Instagram.

Descrição

O que esta skill faz

Analyzes public sentiment for brands and products across Twitter, Reddit, and Instagram. Pulls up to 70K posts per query, exports bulk CSVs, and classifies them using Python/pandas. Covers 1.5B+ indexed posts.

Access to 1.5B+ indexed posts with bulk CSV export lets you analyze tens of thousands of posts in a single query, far beyond what manual social browsing or smaller monitoring tools support.

Quando usá-la

  • Tracking brand reputation after a product launch
  • Detecting PR crises from spikes in negative mentions
  • Comparing competitor sentiment across platforms
  • Identifying recurring complaints buried in customer posts
  • Monitoring marketing campaign reception in real time

Exemplo de fluxo de trabalho

ENTRADA

User asks: Analyze sentiment for Notion on Twitter over the last 30 days

AGENTE
  1. 1

    Calls xpoz.getTwitterPostsByKeywords with query 'Notion' and a 30-day date range

  2. 2

    Polls checkOperationStatus every 5 seconds until the operation completes

  3. 3

    Downloads the bulk CSV using the returned dataDumpExportOperationId

  4. 4

    Loads the CSV with pandas and classifies each post as positive, negative, or neutral using keyword scoring

  5. 5

    Computes an engagement-weighted sentiment score and extracts top themes and viral posts

SAÍDA

Sentiment score 72/100 from 14,832 posts: 58% positive, 24% negative, 18% neutral — top negative theme is Performance (2K posts, 81% negative), top positive theme is UX (1.8K posts, 72% positive)

Requisitos

  • Xpoz account (free tier available) at xpoz.ai — required for OAuth authentication via the xpoz-setup skill

Instalação

clawhub install atyachin/social-sentiment

Execute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.

Links