io.github.LunarPerovskite/docalyze
Read and analyze local documents (PDF, Excel, CSV, Word, PowerPoint, images). No API keys needed.
Installation
Ajoutez ceci à la configuration MCP de votre client (Claude Desktop : claude_desktop_config.json ; Cursor : .cursor/mcp.json ; VS Code et les autres utilisent la même forme « mcpServers »).
docalyze-mcp-serverclaude mcp add docalyze -- uvx docalyze-mcp-server==0.2.1
{
"mcpServers": {
"docalyze": {
"command": "uvx",
"args": [
"docalyze-mcp-server==0.2.1"
]
}
}
}
docalyze-mcp-serverclaude mcp add docalyze -- npx -y docalyze-mcp-server@0.2.1
{
"mcpServers": {
"docalyze": {
"command": "npx",
"args": [
"-y",
"docalyze-mcp-server@0.2.1"
]
}
}
}
Documentation
Tirée du README du projet sur GitHub, actualisée chaque semaine. Les liens ouvrent GitHub.
Sommaire
Docalyze MCP Server
An MCP (Model Context Protocol) server that lets AI assistants read and visually analyze local documents — PDFs, Excel spreadsheets, CSV files, Word documents, PowerPoint presentations, and images.
No API keys required. The host AI (GitHub Copilot, Claude, etc.) does all the reasoning directly.
Supported Formats
| Format | Extensions | Read | Visual |
|---|---|---|---|
.pdf | ✅ | ✅ | |
| Excel | .xlsx, .xls | ✅ | ✅ |
| CSV / TSV | .csv, .tsv | ✅ | — |
| JSON | .json | ✅ | — |
| Word | .docx | ✅ | ✅ |
| PowerPoint | .pptx | ✅ | ✅ |
| Plain text | .txt, .md | ✅ | — |
| Images | .png, .jpg, .jpeg, .gif, .bmp, .tiff, .webp | — | ✅ |
Tools
| Tool | Description |
|---|---|
list_documents | List files under a directory, filtered by glob pattern |
document_info | Get metadata (size, modified date, sheets) for a file |
read_document | Extract text content from a document with pagination |
visual_evaluate_document | Return page images inline so the AI can analyze charts, tables, and diagrams |
Installation
From VS Code (recommended)
Search for docalyze in the MCP server gallery (Extensions sidebar → MCP tab) and click Install.
From PyPI
pip install docalyze-mcp-server
From npm
npx docalyze-mcp-server
This requires uv or pipx installed — the npm wrapper calls uvx to run the Python package automatically.
Manual setup
Add to your VS Code mcp.json (or settings.json):
{
"servers": {
"docalyze": {
"type": "stdio",
"command": "python",
"args": ["-m", "docalyze_mcp_server"],
"env": {
"PYTHONIOENCODING": "utf-8"
}
}
}
}
Or, if you installed via pip and want to use the entry point:
{
"servers": {
"docalyze": {
"type": "stdio",
"command": "docalyze-mcp-server"
}
}
}
Optional Dependencies
The base install handles PDF, Excel, CSV, JSON, and plain text. For additional formats:
# Word documents
pip install docalyze-mcp-server[docx]
# PowerPoint
pip install docalyze-mcp-server[pptx]
# OCR (requires Tesseract installed on your system)
pip install docalyze-mcp-server[ocr]
# Everything
pip install docalyze-mcp-server[all]
Configuration
The server reads documents from a configurable root directory. Set the DOCUMENTS_ROOT environment variable to change it:
{
"servers": {
"docalyze": {
"type": "stdio",
"command": "docalyze-mcp-server",
"env": {
"DOCUMENTS_ROOT": "/path/to/your/documents"
}
}
}
}
If not set, it defaults to the directory containing the server script.
License
MIT
Questions fréquentes
Qu'est-ce que io.github.LunarPerovskite/docalyze ?
Read and analyze local documents (PDF, Excel, CSV, Word, PowerPoint, images). No API keys needed.
Comment installer io.github.LunarPerovskite/docalyze ?
Il s'exécute sur votre machine. Dans Claude Code : claude mcp add docalyze -- uvx docalyze-mcp-server==0.2.1. Pour les autres clients, collez la configuration JSON de cette page dans leurs réglages MCP.
Faut-il une clé API pour io.github.LunarPerovskite/docalyze ?
Son entrée au registre n'indique aucune clé ni secret obligatoire. Certaines fonctions peuvent tout de même demander un compte auprès du service auquel il se connecte.
io.github.LunarPerovskite/docalyze est-il open source ?
Oui. Son code est sur GitHub sous licence MIT.
Qui maintient io.github.LunarPerovskite/docalyze ?
Il est publié par LunarPerovskite. AI.tn le référence sans le maintenir ni le recommander.
Issu du registre MCP officiel. Être listé n'est pas une recommandation : vérifiez le code et les permissions d'un serveur avant de lui donner accès à vos données.