Audiense Insights

官方

來自Audiense報告的行銷洞察與受眾分析，涵蓋人口統計、文化、影響者及內容互動分析。

GitHub
19

你可以用 Audiense Insights MCP 做什麼？

  • 列出您的 Audiense 報告 — 要求助理使用 get-reports 擷取所有可用的洞察報告。
  • 摘要完整報告 — 透過 report-summary 請求結構化的區隔、頂尖洞察與影響者概覽。
  • 取得受眾人口統計與心理特徵 — 使用 get-audience-insights 提取年齡、興趣與人格等彙總特質。
  • 比較影響者與基準線 — 使用 compare-audience-influencers 識別受眾中獨特且高關聯度的影響者。
  • 分析內容互動 — 使用 get-audience-content 探索受眾的熱門貼文、主題標籤、網域與分享連結。

文件

⚠️ 已棄用

🚫 此儲存庫不再維護。

Audiense Insights MCP 已遷移至遠端模型。如需了解如何使用新的遠端 MCP，請透過 support@audiense.com 與我們聯繫。

🏆 Audiense Insights MCP 伺服器

此伺服器基於 Model Context Protocol (MCP)，可讓 Claude 或任何其他相容 MCP 的用戶端與您的 Audiense Insights 帳戶互動。它能從 Audiense 報告中提取行銷洞察與受眾分析，涵蓋人口統計、文化、影響者及內容互動分析。

🚀 先決條件

使用此伺服器前，請確保您具備：

  • Node.js（v18 或更高版本）
  • Claude 桌面應用程式
  • Audiense Insights 帳戶及 API 憑證
  • X/Twitter API Bearer Token（選用，用於豐富的影響者資料）

⚙️ 設定 Claude 桌面版

  1. 開啟 Claude 桌面版的設定檔：

    • MacOS： 
      code ~/Library/Application\ Support/Claude/claude_desktop_config.json
    • Windows： 
      code %AppData%\Claude\claude_desktop_config.json

  2. 新增或更新以下設定：

    "mcpServers": {
  "audiense-insights": {
    "command": "npx",
    "args": [
     "-y",
      "mcp-audiense-insights"
    ],
    "env": {
      "AUDIENSE_CLIENT_ID": "your_client_id_here",
      "AUDIENSE_CLIENT_SECRET": "your_client_secret_here",
      "TWITTER_BEARER_TOKEN": "your_token_here"
    }          
  }     
}

  3. Save the file and restart Claude Desktop.

🛠️ Available Tools

📌 get-reports

Description: Retrieves the list of Audiense insights reports owned by the authenticated user.

  • Parameters: None
  • Response:
    • List of reports in JSON format.

📌 get-report-info

Description: Fetches detailed information about a specific intelligence report, including:

  • Status

  • Segmentation type

  • Audience size

  • Segments

  • Access links

  • Parameters:

    • report_id (string): The ID of the intelligence report.

  • Response:

    • Full report details in JSON format.
    • If the report is still processing, returns a message indicating the pending status.

📌 get-audience-insights

Description: Retrieves aggregated insights for a given audience, including:

  • Demographics: Gender, age, country.

  • Behavioral traits: Active hours, platform usage.

  • Psychographics: Personality traits, interests.

  • Socioeconomic factors: Income, education status.

  • Parameters:

    • audience_insights_id (string): The ID of the audience insights.
    • insights (array of strings, optional): List of specific insight names to filter.

  • Response:

    • Insights formatted as a structured text list.

📌 get-baselines

Description: Retrieves available baseline audiences, optionally filtered by country.

  • Parameters:

    • country (string, optional): ISO country code to filter by.

  • Response:

    • List of baseline audiences in JSON format.

📌 get-categories

Description: Retrieves the list of available affinity categories that can be used in influencer comparisons.

  • Parameters: None
  • Response:
    • List of categories in JSON format.

📌 compare-audience-influencers

Description: Compares influencers of a given audience with a baseline audience. The baseline is determined as follows:

  • If a single country represents more than 50% of the audience, that country is used as the baseline.
  • Otherwise, the global baseline is used.
  • If a specific segment is selected, the full audience is used as the baseline.

Each influencer comparison includes:

  • Affinity (%) – How well the influencer aligns with the audience.

  • Baseline Affinity (%) – The influencer’s affinity within the baseline audience.

  • Uniqueness Score – How distinct the influencer is compared to the baseline.

  • Parameters:

    • audience_influencers_id (string): ID of the audience influencers.
    • baseline_audience_influencers_id (string): ID of the baseline audience influencers.
    • cursor (number, optional): Pagination cursor.
    • count (number, optional): Number of items per page (default: 200).
    • bio_keyword (string, optional): Filter influencers by bio keyword.
    • entity_type (enum: person | brand, optional): Filter by entity type.
    • followers_min (number, optional): Minimum number of followers.
    • followers_max (number, optional): Maximum number of followers.
    • categories (array of strings, optional): Filter influencers by categories.
    • countries (array of strings, optional): Filter influencers by country ISO codes.

  • Response:

    • List of influencers with affinity scores, baseline comparison, and uniqueness scores in JSON format.

📌 get-audience-content

Description: Retrieves audience content engagement details, including:

  • Liked Content: Most popular posts, domains, emojis, hashtags, links, media, and a word cloud.
  • Shared Content: Most shared content categorized similarly.
  • Influential Content: Content from influential accounts.

Each category contains:

  • popularPost: Most engaged posts.

  • topDomains: Most mentioned domains.

  • topEmojis: Most used emojis.

  • topHashtags: Most used hashtags.

  • topLinks: Most shared links.

  • topMedia: Shared media.

  • wordcloud: Most frequently used words.

  • Parameters:

    • audience_content_id (string): The ID of the audience content.

  • Response:

    • Content engagement data in JSON format.

📌 report-summary

Description: Generates a comprehensive summary of an Audiense report, including:

  • Report metadata (title, segmentation type)

  • Full audience size

  • Detailed segment information

  • Top insights for each segment (bio keywords, demographics, interests)

  • Top influencers for each segment with comparison metrics

  • Parameters:

    • report_id (string): The ID of the intelligence report to summarize.

  • Response:

    • Complete report summary in JSON format with structured data for each segment
    • For pending reports: Status message indicating the report is still processing
    • For reports without segments: Message indicating there are no segments to analyze

💡 Predefined Prompts

This server includes a preconfigured prompts

  • audiense-demo: Helps analyze Audiense reports interactively.
  • segment-matching: A prompt to match and compare audience segments across Audiense reports, identifying similarities, unique traits, and key insights based on demographics, interests, influencers, and engagement patterns.

Usage:

  • Accepts a reportName argument to find the most relevant report.
  • If an ID is provided, it searches by report ID instead.

Use case: Structured guidance for audience analysis.

🛠️ Troubleshooting

Tools Not Appearing in Claude

  1. Check Claude Desktop logs:
tail -f ~/Library/Logs/Claude/mcp*.log
  1. Verify environment variables are set correctly.
  2. Ensure the absolute path to index.js is correct.

Authentication Issues

  • Double-check OAuth credentials.
  • Ensure the refresh token is still valid.
  • Verify that the required API scopes are enabled.

📜 Viewing Logs

To check server logs:

For MacOS/Linux:

tail -n 20 -f ~/Library/Logs/Claude/mcp*.log

For Windows:

Get-Content -Path "$env:AppData\Claude\Logs\mcp*.log" -Wait -Tail 20

🔐 安全性考量

  • 妥善保管 API 憑證 – 切勿在公開儲存庫中暴露。
  • 使用環境變數來管理敏感資料。

📄 授權條款

本專案採用 Apache 2.0 授權條款。詳情請參閱 LICENSE 檔案。