Claude MCP Data Explorer
Claude MCP データエクスプローラー(Windows 用)
Claude によるデータ探索のための Model Context Protocol (MCP) サーバーの TypeScript 実装です。このサーバーは Claude Desktop と統合されており、CSV ファイルの読み込みや JavaScript データ分析スクリプトの実行ツールを提供することで、高度なデータ分析を可能にします。
前提条件
Node.js v16+ - Node.js をダウンロード
Claude Desktop - Claude Desktop をダウンロード
Related MCP server: mcp-csv-analyst
インストール(Windows 用に更新)
このリポジトリをクローンする
git clone https://github.com/tofunori/claude-mcp-data-explorer.git cd claude-mcp-data-explorer依存関係をインストールする
npm installセットアップスクリプトをビルドして実行する
npm run setupこれにより、次のようになります。
TypeScriptコードをJavaScriptにビルドする
コンパイルされたJavaScriptを使用するようにClaude Desktopを設定する
必要なディレクトリを作成する
Claude Desktopを再起動して開発者モードを有効にします
Claude Desktopを完全に閉じる
Claudeデスクトップを起動
ヘルプ→開発者モードを有効にする
手動テスト
次のコマンドを実行してサーバーを直接テストできます。
npm run build
npm run startサーバーはエラーなく起動するはずです。正常に実行できれば、Claude Desktop でもサーバーを使用できるはずです。
仕組み
この MCP サーバーは、Claude に 2 つの主要なツールを提供します。
load-csv - 分析のためにCSVデータをメモリにロードします
run-script - データ処理と分析のためのJavaScriptコードを実行します
また、構造化されたデータ探索プロセスを通じて Claude をガイドするプロンプト テンプレートも含まれています。
使用法
Claudeデスクトップを起動
「データの探索」プロンプトテンプレートを選択します
このプロンプトはセットアップ後にClaude Desktopに表示されます
CSVファイルのパスと探索トピックを入力してください
ファイルパスの例:
C:/Users/YourName/Documents/data.csvトピック例:「地域別の売上動向」
クロードにデータを分析してもらいましょう
クロードはCSVファイルを読み込み、自動的にインサイトを生成します
サーバーはチャンク化を使用して大きなファイルを効率的に処理します
トラブルシューティング
クロードはMCPサーバーを表示しません
%APPDATA%\Claude\claude_desktop_config.jsonにある構成ファイルを確認します。これは、distディレクトリ内のコンパイルされたJavaScriptファイルを指す必要があります。
npm run buildでプロジェクトを再構築してみてください開発者モードを有効にして、MCP ログファイルを確認します (開発者 → MCP ログファイルを開く)
開発者 → すべてのMCPサーバーを再読み込みを使用して強制的に更新します
ファイルの読み取り権限エラー
クロードがCSVファイルの場所にアクセスできることを確認してください
スラッシュ(
/)またはエスケープされたバックスラッシュ(\\)を使用した絶対パスを使用してみてください
スクリプト内のJavaScriptエラー
スクリプトが許可されたモジュールと互換性があることを確認してください
クロードの回答にあるエラーメッセージを確認します
ライセンス
MIT ライセンス - 詳細については LICENSE ファイルを参照してください。
謝辞
Anthropicの公式MCP TypeScript SDKに基づいています
例とインスピレーションを提供してくれたMCPコミュニティに感謝します
Available Tools
2 toolsload-csvC
Load a CSV file into a DataFrame for analysis
| Name | Required | Description | Default |
|---|---|---|---|
| csv_path | Yes | Path to the CSV file to load | |
| df_name | No | Name for the DataFrame (optional, defaults to df_1, df_2, etc.) |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states the action 'load a CSV file into a DataFrame' but lacks details on permissions needed, error handling (e.g., invalid paths), rate limits, or what happens after loading (e.g., memory usage, persistence). For a tool with no annotation coverage, this is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, efficient sentence that front-loads the core action ('Load a CSV file') and purpose ('for analysis'). There is zero waste, making it highly concise and well-structured for quick understanding.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no annotations, no output schema, and a tool that performs data loading (a potentially complex operation with file I/O), the description is incomplete. It doesn't cover behavioral aspects like error conditions, return values, or dependencies, leaving gaps for an AI agent to use it correctly in varied contexts.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents both parameters (csv_path and df_name) with clear descriptions. The description adds no additional meaning beyond implying CSV loading for analysis, which aligns with the schema but doesn't provide extra syntax or format details. Baseline 3 is appropriate when the schema does the heavy lifting.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the verb 'load' and resource 'CSV file', specifying it's for analysis via a DataFrame. It distinguishes from the sibling 'run-script' by focusing on data loading rather than script execution. However, it doesn't explicitly differentiate from potential other data loading tools (none listed), keeping it at 4.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives. The description implies it's for loading CSV files into DataFrames, but there's no mention of prerequisites (e.g., file accessibility), when not to use it (e.g., for non-CSV files), or alternatives like 'run-script' for other data processing. This leaves usage context vague.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
run-scriptC
Execute a JavaScript script for data analysis and visualization
| Name | Required | Description | Default |
|---|---|---|---|
| script | Yes | JavaScript script to execute |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states the tool executes a JavaScript script but doesn't describe safety aspects (e.g., sandboxing, permissions), performance traits (e.g., execution time limits, resource usage), or what happens upon execution (e.g., output format, side effects). For a tool that runs arbitrary code with no annotation coverage, this is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, efficient sentence that directly states the tool's purpose without unnecessary words. It is appropriately sized and front-loaded, with every part contributing essential information (verb, resource, domain).
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of executing arbitrary JavaScript code, the lack of annotations, and no output schema, the description is incomplete. It doesn't address critical aspects like security implications, error handling, or what the tool returns (e.g., visualization output, analysis results). For a tool with such potential impact, more context is needed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 100% description coverage, with one parameter 'script' fully documented in the schema. The description adds no additional meaning about parameters beyond what the schema provides (e.g., no examples of script content, no constraints on JavaScript features). With high schema coverage, the baseline score of 3 is appropriate as the description doesn't compensate but doesn't detract either.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose with a specific verb ('Execute') and resource ('JavaScript script'), and specifies the domain ('for data analysis and visualization'). It doesn't distinguish from the sibling tool 'load-csv', which appears to be a different operation, so it doesn't explicitly differentiate from siblings.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. It mentions the domain (data analysis and visualization) but doesn't specify prerequisites, limitations, or when not to use it. There's no explicit comparison with the sibling tool 'load-csv' or other potential tools.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
2 tool updates
- First observed
load-csv - First observed
run-script
TDQS
Scored across 2 tools
The two tools have clearly distinct purposes: one loads CSV data into a DataFrame, while the other executes JavaScript scripts for analysis and visualization. There is no overlap or ambiguity between these functions, making it easy for an agent to select the correct tool.
Both tools use a verb-noun pattern (load-csv, run-script), which is consistent and readable. The hyphenated style is maintained throughout, though the specific convention (hyphens vs. underscores) is less important than the consistency, which is good here with only minor deviations from common patterns.
With only two tools, the server feels thin for a 'Data Explorer' purpose, as it lacks essential operations like data querying, filtering, transformation, or exporting. While the tools are functional, the count is too low to adequately cover the expected scope of data exploration and analysis.
For a data exploration server, there are significant gaps: no tools for querying data, filtering, aggregating, visualizing beyond scripts, or exporting results. The surface is severely incomplete, as agents cannot perform basic data exploration tasks without relying heavily on external scripts, leading to potential failures in common workflows.
Maintenance
Related MCP Connectors
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Query, join, profile, clean and convert CSV/JSON/Parquet with server-side DuckDB over MCP.
Query your warehouse or a CSV with Claude/ChatGPT over MCP, governed by table-level ACL + audit.
Augments MCP Server - A comprehensive framework documentation provider for Claude Code
Related MCP Servers
- -licenseNot gradedqualityNot gradedmaintenanceA Model Context Protocol server that analyzes customer cancellation survey data, enabling Claude AI integration through HTTP/SSE transport.-
- AlicenseAqualityNot gradedmaintenanceAn MCP server that enables AI assistants to load, query, and analyze local CSV files using tools for filtering, aggregation, and grouping. It provides capabilities to describe schemas, calculate statistics, and sample data directly from CSV files.6-
- FlicenseNot gradedqualityDmaintenanceA local MCP server for analyzing CSV files from your filesystem, particularly suited for chatbot conversation logs. Allows listing, reading, filtering, merging, and statistical analysis of CSV data via natural language.-
- FlicenseAqualityDmaintenanceAn MCP server for dataset exploration and analysis, enabling LLM clients to perform summary, correlation, distribution, missing value analysis, data cleaning, and statistical tests directly on CSV files.3-