MCP SERVER

Disbursement MCP Server

A knowledge graph–powered assistant for the DisburseLoan domain. Connect it to your AI tool to query validations, flows, entities, and more.

Service Status

MCP Server checking…
Neo4j checking…
Tools —

Server URL

http://35-234-220-25.sslip.io/sse

You need an API key to connect — contact the administrator. Every client must send it as: Authorization: Bearer <your-key>

Supported Clients

Claude Code Claude.ai Web Cursor Continue.dev Windsurf Zed Codex CLI Gemini CLI ChatGPT Web Gemini Web

MCP is an open protocol — CLI tools from OpenAI (Codex) and Google (Gemini) support it. Their web interfaces do not.

Setup — Claude Code (CLI / Desktop / VS Code / JetBrains)

1

Open your settings file

Edit .claude/settings.json (global) or .claude/settings.local.json (project-only).

2

Add the server block

{
  "mcpServers": {
    "disbursement-mcp": {
      "url": "http://35-234-220-25.sslip.io/sse",
      "headers": {
        "Authorization": "Bearer <your-api-key>"
      }
    }
  }
}
3

Restart Claude Code

The MCP tools appear automatically in your next session.

Setup — Claude.ai (Web)

1

Go to Settings → Integrations

Open claude.ai → profile → Settings → Integrations → Add Integration.

2

Enter the server details

  • URL: http://35-234-220-25.sslip.io/sse
  • Header name: Authorization
  • Header value: Bearer <your-api-key>
3

Enable in a conversation

New chat → click the Tools icon → toggle disbursement-mcp on.

Claude.ai web requires HTTPS. If the URL starts with http://, it will not work here — ask the admin to enable HTTPS first.

Setup — Local AI Tools

These editors run locally on your machine and support MCP servers over HTTP/SSE.

Cursor

AI code editor — cursor.com

Go to Cursor Settings → MCP (or edit ~/.cursor/mcp.json):

{
  "mcpServers": {
    "disbursement-mcp": {
      "url": "http://35-234-220-25.sslip.io/sse",
      "headers": {
        "Authorization":
          "Bearer <key>"
      }
    }
  }
}

Restart Cursor — tools appear in Agent mode.

Continue.dev

VS Code / JetBrains extension — continue.dev

Edit ~/.continue/config.json:

{
  "mcpServers": [
    {
      "name": "disbursement-mcp",
      "transport": {
        "type": "sse",
        "url": "http://35-234-220-25.sslip.io/sse",
        "headers": {
          "Authorization":
            "Bearer <key>"
        }
      }
    }
  ]
}

Windsurf

AI IDE by Codeium — codeium.com/windsurf

Go to Windsurf Settings → MCP Servers → Add Server → choose SSE:

  • URL: http://35-234-220-25.sslip.io/sse
  • Header: Authorization: Bearer <key>

Or edit ~/.codeium/windsurf/mcp_config.json with the same JSON format as Cursor.

Zed

Multiplayer code editor — zed.dev

Edit ~/.config/zed/settings.json:

{
  "context_servers": {
    "disbursement-mcp": {
      "settings": {
        "url": "http://35-234-220-25.sslip.io/sse",
        "headers": {
          "Authorization":
            "Bearer <key>"
        }
      }
    }
  }
}

OpenAI Codex CLI

OpenAI's coding CLI — github.com/openai/codex

Edit ~/.codex/config.toml:

# ~/.codex/config.toml
[[mcp_servers]]
name    = "disbursement-mcp"
url     = "http://35-234-220-25.sslip.io/sse"

[mcp_servers.headers]
Authorization = "Bearer <key>"

Then run: codex --mcp disbursement-mcp

Gemini CLI

Google's coding CLI — github.com/google-gemini/gemini-cli

Edit ~/.gemini/settings.json:

{
  "mcpServers": {
    "disbursement-mcp": {
      "url": "http://35-234-220-25.sslip.io/sse",
      "headers": {
        "Authorization":
          "Bearer <key>"
      }
    }
  }
}

Then run: gemini — tools are available in the session.

For all local tools — after saving the config, fully restart the editor or CLI. MCP tools are available in the AI agent/chat panel. ChatGPT web and Gemini web do not support MCP — use their CLI versions (Codex CLI / Gemini CLI) instead.

Using with Local LLM Providers (Ollama / LM Studio)

Ollama and LM Studio are model runners — they serve AI models locally but have no MCP client built in. You connect them to an MCP-capable editor which handles the MCP side.

How it works: Editor (MCP client) ↔ This MCP Server  +  Editor ↔ Ollama/LM Studio (local model)

Continue.dev + Ollama

Best combo for local LLMs — free & open source

Step 1 — Install Ollama and pull a model:

curl -fsSL https://ollama.com/install.sh | sh
ollama pull qwen2.5-coder:7b

Step 2 — Edit ~/.continue/config.json:

{
  "models": [{
    "provider": "ollama",
    "model": "qwen2.5-coder:7b",
    "title": "Local Qwen"
  }],
  "mcpServers": [{
    "name": "disbursement-mcp",
    "transport": {
      "type": "sse",
      "url": "http://35-234-220-25.sslip.io/sse",
      "headers": {
        "Authorization": "Bearer <key>"
      }
    }
  }]
}

Continue.dev + LM Studio

GUI-based local model manager — lmstudio.ai

Step 1 — Open LM Studio → load a model → start the local server (port 1234).

Step 2 — Edit ~/.continue/config.json:

{
  "models": [{
    "provider": "lmstudio",
    "model": "AUTODETECT",
    "title": "LM Studio"
  }],
  "mcpServers": [{
    "name": "disbursement-mcp",
    "transport": {
      "type": "sse",
      "url": "http://35-234-220-25.sslip.io/sse",
      "headers": {
        "Authorization": "Bearer <key>"
      }
    }
  }]
}

Cursor + Ollama

Use Ollama as the model inside Cursor

Step 1 — In Cursor Settings → Models → add a custom model pointing to http://localhost:11434 (Ollama's OpenAI-compatible endpoint).

Step 2 — MCP config is the same as the Cursor section above — Cursor handles both the local model and MCP independently.

Recommended Local Models

Models with good tool/function-calling support
  • qwen2.5-coder:7b — best for code, fast
  • qwen2.5-coder:32b — best quality local
  • llama3.1:8b — general purpose
  • mistral:7b — lightweight option

Tool calling quality matters for MCP — models without function-calling support will not invoke MCP tools correctly.

Available Tools

Loading tools…

Example Prompts

The API key is not included in this guide — contact the server administrator to obtain yours.