File2markdown MCP server
{
"mcpServers": {
"file2markdown": {
"type": "http",
"url": "https://mcp.file2markdown.ai/mcp"
}
}
}{
"mcpServers": {
"file2markdown": {
"type": "http",
"url": "https://mcp.file2markdown.ai/mcp"
}
}
}{
"mcpServers": {
"file2markdown": {
"type": "http",
"url": "https://mcp.file2markdown.ai/mcp"
}
}
}{
"servers": {
"file2markdown": {
"type": "http",
"url": "https://mcp.file2markdown.ai/mcp"
}
}
}{
"mcpServers": {
"file2markdown": {
"type": "http",
"url": "https://mcp.file2markdown.ai/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.0
What it does
file2markdown MCP server converts documents and web pages to Markdown via a hosted HTTP endpoint. It supports formats that LLM clients cannot parse natively, including PDF, DOCX, XLSX, PPTX, EPUB, and images with OCR. It is intended for developers and AI agent workflows using Claude, Cursor, or any MCP-compatible client.
- Convert a publicly accessible PDF or Office document URL to Markdown for use in an LLM prompt.
- Extract structured text from scanned PDFs or images using OCR (Pro tier) and receive clean Markdown output.
- Check remaining daily conversion quota and supported format details before building an automated document-processing workflow.
Tools this server exposes
| Tool | Does |
|---|---|
| convert_url | Fetches a public URL (web page, PDF, Office document, etc.) and returns its content as Markdown. |
| convert_base64 | Converts file contents supplied as base64-encoded data directly to Markdown, for programmatic clients. |
| list_supported_formats | Returns the list of supported input formats along with per-tier limits and which features require a Pro key. |
| usage_status | Returns the caller's current tier and the number of conversions remaining for the day. |
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.