sapient-mcp

Aproduction-ready Model Context Protocol (MCP) serverthat enables LLMs (Claude, Copilot, Cursor, etc.) to automateSAP GUIusing theRoboSAPienslibrary.

# Clone the repo git clone https://github.com/yourorg/sapient-mcp.git cd sapient-mcp # Install (using pip or uv) pip install -e . # Or with uv (recommended) uv pip install -e .

{ "mcpServers": { "sapient": { "command": "python", "args": ["-m", "sapient_mcp"], "env": { "SAPIENT_MCP_SAPLOGON_PATH": "C:\\Program Files (x86)\\SAP\\FrontEnd\\SAPgui\\saplogon.exe", "SAPIENT_MCP_CAPS": "screenshot,codegen,advanced", "SAPIENT_MCP_OUTPUT_DIR": "C:\\sapient_output" } } } }

claude mcp add sapient python -m sapient_mcp \ -- --caps screenshot,codegen,advanced

Or edit~/.claude/mcp.json/.claude/mcp.jsonin your project root:

{ "mcpServers": { "sapient": { "command": "python", "args": ["-m", "sapient_mcp", "--caps", "screenshot,codegen,advanced"], "env": { "SAPIENT_MCP_SAPLOGON_PATH": "C:\\Program Files (x86)\\SAP\\FrontEnd\\SAPgui\\saplogon.exe", "SAPIENT_MCP_OUTPUT_DIR": "C:\\sapient_output" } } } }

Go toCursor Settings→MCP→Add new MCP Server→ typecommand, then enter:

python -m sapient_mcp --caps screenshot,codegen,advanced

{ "mcpServers": { "sapient": { "command": "python", "args": ["-m", "sapient_mcp", "--caps", "screenshot,codegen,advanced"], "env": { "SAPIENT_MCP_SAPLOGON_PATH": "C:\\Program Files (x86)\\SAP\\FrontEnd\\SAPgui\\saplogon.exe", "SAPIENT_MCP_OUTPUT_DIR": "C:\\sapient_output" } } } }

code --add-mcp '{"name":"sapient","command":"python","args":["-m","sapient_mcp","--caps","screenshot,codegen,advanced"]}'

Or add to.vscode/mcp.jsonin your workspace:

{ "servers": { "sapient": { "type": "stdio", "command": "python", "args": ["-m", "sapient_mcp", "--caps", "screenshot,codegen,advanced"], "env": { "SAPIENT_MCP_SAPLOGON_PATH": "C:\\Program Files (x86)\\SAP\\FrontEnd\\SAPgui\\saplogon.exe", "SAPIENT_MCP_OUTPUT_DIR": "C:\\sapient_output" } } } }

Edit~/.codeium/windsurf/mcp_config.json:

{ "mcpServers": { "sapient": { "command": "python", "args": ["-m", "sapient_mcp", "--caps", "screenshot,codegen,advanced"], "env": { "SAPIENT_MCP_SAPLOGON_PATH": "C:\\Program Files (x86)\\SAP\\FrontEnd\\SAPgui\\saplogon.exe", "SAPIENT_MCP_OUTPUT_DIR": "C:\\sapient_output" } } } }

Note:In stdio mode, SAPient MCPnever writes to stdout or stderr— all logs go to the log file only. This is required for clean JSON-RPC communication across all clients.

Option B — SSE/HTTP mode (remote / CI / multi-user)

Best when the SAP machine is separate from where the AI client runs, or when multiple developers want to share one SAPient instance.

Run the server on your Windows SAP machine:

python -m sapient_mcp --port 8765 --caps screenshot,codegen,advanced

Then pointany MCP clientat the HTTP endpoint:

{ "mcpServers": { "sapient": { "url": "http://sap-windows-machine:8765/mcp" } } }

This works identically in Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, and any other MCP-compatible client.

python -m sapient_mcp --config C:\sap_config.json

All settings can be provided as env vars with theROBOSAP_MCP_prefix:

You:Create a purchase order for vendor 100001, company code 1000, with 50 units of material ABC-001
- Callsap_execute_transaction("/nME21N")
- Callsap_get_window_title()→ "Create Purchase Order"
- Callsap_fill_text_field("Vendor", "100001")
- Callsap_fill_text_field("Company Code", "1000")
- Callsap_send_key("Enter")to accept header
- Callsap_activate_tab("Item Overview")
- Callsap_fill_cell("1", "Material", "ABC-001")
- Callsap_fill_cell("1", "Quantity", "50")
- Callsap_push_button("Save")
- Callsap_read_status_bar()→ "Purchase order 4500001234 created"

sapient-mcp/ ├── pyproject.toml ├── README.md ├── sap_config.json # Example config file ├── claude_desktop_config.json # Example Claude Desktop config └── src/ └── sapient_mcp/ ├── __init__.py ├── __main__.py # Entry point + CLI arg parsing ├── config.py # Pydantic settings (env/file/CLI) ├── session.py # SAPSessionManager singleton └── server.py # FastMCP server + all 27 tool definitions

Logs are written to{output_dir}/sapient_mcp.log.

Instdio mode, logs never go to stdout/stderr (would break JSON-RPC). InSSE mode, logs also appear on stderr.

D2C eCommerce fulfillment platform: manage orders, inventory, shipments, campaigns, and billing via AI agents

The Company Formation MCP for AI coding tools.

MCP server for Fiken — Norway's popular accounting software. Manage invoices, contacts, journal entries, and bank accounts via AI agents.

MCP server for 1C:Enterprise (ERP, Accounting, etc.) — 51 tools for metadata, documents, registers, reports and more

KLAVE is an MCP server that gives Claude the ability to autonomously + privately negotiate any deal on your behalf...no human back-and-forth required.

Agentic pipeline that transforms ideas to revenue — for solo founders and bootstrappers.

Open protocol for professional service delivery. AI agents can discover, schedule, verify and settle professional services.

Run MCP tools in production without managing your own server — built-in retries, permissions, and observability.

Official Airtable MCP server and skills for working with bases, records, workflows, and business operations from AI agents.

What Shopify did for ecommerce, Chipp does for AI agents. Build, deploy, and monetize AI agents for your business — no engineering team required.