mdkit

# Mdkit

**Convert documents (PDF, Office, HTML) into clean, LLM-ready Markdown and JSON.**

mdkit converts documents — PDF, Word, PowerPoint, Excel and HTML — into clean, LLM-ready Markdown and JSON over a simple REST API and an MCP server, so developers can feed real-world documents straight into a model or a RAG pipeline without hand-cleaning them. Small files convert synchronously in one call; larger jobs run through an async job queue with polling and signed webhooks. Free tier to try it, metered credits for volume. An alternative to Jina Reader, Firecrawl, Docling, MarkItDown and Unstructured.

## Links

- **Website:** https://mdkit.online/go/github
- **API docs:** https://api.mdkit.online/docs · [OpenAPI](https://api.mdkit.online/openapi.json)
- **Pricing:** https://mdkit.online/pricing
- **Convert HTML free:** https://mdkit.online/tools/html-to-markdown
- **llms.txt:** https://mdkit.online/llms.txt

## MCP server

Streamable-HTTP endpoint: `https://api.mdkit.online/mcp/` — the anonymous
tier works without a key; an API key raises the limits.

Claude Code:

```sh
claude mcp add --transport http mdkit https://api.mdkit.online/mcp/
```

Cursor / Windsurf / Cline / Claude Desktop:

```json
{
  "mcpServers": {
    "mdkit": {
      "url": "https://api.mdkit.online/mcp/"
    }
  }
}
```

One-click installs for every client: https://mdkit.online/connect

## API

Base URL `https://api.mdkit.online/` — authenticate with an `X-API-Key` header;
anonymous calls work at a lower rate limit.