Gemina MCP server
{
"mcpServers": {
"gemina": {
"type": "http",
"url": "https://api.gemina.co/api/v1/mcp/"
}
}
}{
"mcpServers": {
"gemina": {
"type": "http",
"url": "https://api.gemina.co/api/v1/mcp/"
}
}
}{
"mcpServers": {
"gemina": {
"type": "http",
"url": "https://api.gemina.co/api/v1/mcp/"
}
}
}{
"servers": {
"gemina": {
"type": "http",
"url": "https://api.gemina.co/api/v1/mcp/"
}
}
}{
"mcpServers": {
"gemina": {
"type": "http",
"url": "https://api.gemina.co/api/v1/mcp/"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 2.1.0
What it does
Gemina is an MCP server that tags, extracts data from, and searches documents including invoices, receipts, contracts, and forms. It offers a free tier for tagging (1,500 tags per month) and paid extraction and search features, with authentication via OAuth or API key.
- Tag PDFs and images to automatically extract metadata like vendor name, date, and amount, then rename files based on suggested patterns.
- Extract structured data from invoices, contracts, or custom document templates to pull out specific fields like line items, totals, or custom fields you define.
- Search and aggregate across your indexed documents to find invoices by vendor or date, and calculate sums, averages, and counts grouped by vendor, currency, or time period.
Tools this server exposes
| Tool | Does |
|---|---|
| files_create_upload | Upload a document (PDF or image up to 50 MB) for processing. |
| tag_file | Tag an uploaded document to extract metadata and generate suggested filenames. |
| tag_url | Tag a document from a URL to extract metadata and generate suggested filenames. |
| files_create_extraction_upload | Upload a document for extraction using OCR or predefined models. |
| extract_document | Extract fields from a document using OCR, invoice models, or a custom template you define. |
| get_extraction_result | Retrieve the result of a document extraction. |
| list_extractions | List all extractions for your account. |
| get_extraction | Get details of a specific extraction. |
| get_document | Get details of a document. |
| submit_extraction_feedback | Send corrections to extracted fields to improve future extractions. |
| query_documents | Search indexed documents by vendor, date, amount, type, or free text. |
| aggregate_documents | Calculate sums, averages, and counts of indexed documents grouped by vendor, currency, type, or time period. |
| index_document | Index a document for search and aggregation. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.