Skip to content

Getting Started

ra-mcp is available as a hosted MCP server — no installation needed. Connect it to your favourite AI client and start searching Swedish historical archives immediately.

Server URL:

https://riksarkivet-ra-mcp.hf.space/mcp

Connect to Claude.ai

Claude.ai supports adding MCP servers directly from the web interface.

  1. Open claude.ai and go to a conversation
  2. Click the :material-puzzle: Integrations icon (bottom-left of the chat input)
  3. Select Add more integrations...
  4. Choose Add custom MCP server
  5. Enter:
    • Name: ra-mcp
    • URL: https://riksarkivet-ra-mcp.hf.space/mcp
  6. Click Add

You should now see ra-mcp listed as a connected integration. Try asking:

"Search for trolldom in Swedish historical documents"

Connect to Claude Code (CLI)

Run a single command:

claude mcp add --transport http ra-mcp https://riksarkivet-ra-mcp.hf.space/mcp

Verify the connection:

claude mcp list

Then inside Claude Code, just ask naturally:

"Search for documents mentioning Stockholm from the 1600s"

Connect to Claude Desktop

Add this to your Claude Desktop configuration file:

=== "macOS / Linux"

Edit `~/.config/claude/claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    }
  }
}
```

=== "Windows"

Edit `%APPDATA%\Claude\claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    }
  }
}
```

Restart Claude Desktop after saving the file.

Connect to VS Code / Cursor / Windsurf

Add to your project's .vscode/mcp.json (or global settings):

{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    }
  }
}

Connect to ChatGPT / Other MCP Clients

Any MCP-compatible client can connect using streamable HTTP:

Setting Value
Transport Streamable HTTP
URL https://riksarkivet-ra-mcp.hf.space/mcp

Adding Plugins

ra-mcp can be combined with other MCP servers to create a richer research environment. A common addition is HTR transcription — a server that can transcribe handwritten document images using AI.

Adding the HTR Transcription Plugin

The htrflow-mcp server provides handwritten text recognition (HTR) capabilities, letting your AI assistant transcribe document images directly.

Server URL:

https://riksarkivet-htrflow-mcp.hf.space/gradio_api/mcp/sse

Claude.ai

  1. Open claude.ai → :material-puzzle: Integrations → Add more integrations...
  2. Add custom MCP server
  3. Enter:
    • Name: htrflow-mcp
    • URL: https://riksarkivet-htrflow-mcp.hf.space/gradio_api/mcp/sse
  4. Click Add

Claude Code

claude mcp add --transport sse htrflow-mcp https://riksarkivet-htrflow-mcp.hf.space/gradio_api/mcp/sse

Claude Desktop

Add alongside ra-mcp in your config file:

{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    },
    "htrflow-mcp": {
      "url": "https://riksarkivet-htrflow-mcp.hf.space/gradio_api/mcp/sse"
    }
  }
}

VS Code / Cursor / Windsurf

Add to .vscode/mcp.json:

{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    },
    "htrflow-mcp": {
      "url": "https://riksarkivet-htrflow-mcp.hf.space/gradio_api/mcp/sse"
    }
  }
}

Using Both Servers Together

Once both servers are connected, you can combine search and transcription in a single conversation:

  1. Search for documents using ra-mcp
  2. Browse specific pages to find interesting content
  3. Transcribe document images using htrflow-mcp for pages that need fresh OCR/HTR

"Search for trolldom in court records, then transcribe page 12 of the first result"

Install the CLI

If you want to search and browse directly from your terminal (no AI client needed):

uv pip install ra-mcp

# Search transcribed documents
ra search "trolldom"
ra search "((Stockholm OR Goteborg) AND troll*)"

# Browse specific pages
ra browse "SE/RA/310187/1" --pages "7,8,52" --search-term "trolldom"

Run Your Own Server

For development or private deployments:

# Clone and install
git clone https://github.com/AI-Riksarkivet/ra-mcp.git
cd ra-mcp
uv sync

# Run MCP server on HTTP
uv run ra serve --port 8000

# Then connect to it
claude mcp add --transport http ra-mcp http://localhost:8000/mcp

Or with Docker:

docker run -p 7860:7860 riksarkivet/ra-mcp

Test with MCP Inspector

For debugging or exploring the tools interactively:

cat > mcp.json <<'EOF'
{
  "mcpServers": {
    "ra-mcp": {
      "type": "streamable-http",
      "url": "https://riksarkivet-ra-mcp.hf.space/mcp"
    }
  }
}
EOF

npx -y @modelcontextprotocol/inspector --config ./mcp.json

Open http://localhost:6274 in your browser to use the Inspector UI.