Model Context Protocol (MCP) 2.0

AI Agent Connectivity & Secure Actions

Connect Claude Desktop, Cursor IDE, ChatGPT, and autonomous agents with Future MCA's secure Post-Action layer.

Checking Server...
Actions & Approvals Hub
LEVEL 119 tools

Context & Read Tools

Safe, immediate information retrieval. Agent inspects company master records, active directors, compliance risks, and Companies Act rules.

LEVEL 29 tools

Prepare Action Tools

Creates validated action drafts and requirements checklists (DIR-12, INC-22, SPICe+). NEVER submits directly.

LEVEL 37 tools

Lifecycle & Execution

Requires explicit confirmation, isolated DSC signing, unexpired tokens, and idempotency protection before executing submissions.

Remote MCP Server EndpointStreamable JSON-RPC 2.0

https://mca-portal-ten.vercel.app/api/mcp

STEP-BY-STEP CONNECTION GUIDE IN CLAUDE:
1
Add Connector in Claude
Paste Server URL:
https://mca-portal-ten.vercel.app/api/mcp
2
Select OAuth Options
• Auth: Required when server asks (or Always)
• OAuth Client: Use Anthropic's hosted client metadata
3
Sign In & Authorize
Claude opens popup → select your Workspace → Click Authorize Claude → Complete!
Or add via claude_desktop_config.json:
{
  "mcpServers": {
    "future-mca": {
      "url": "https://mca-portal-ten.vercel.app/api/mcp"
    }
  }
}

Live Interactive MCP Tool Playground

Simulate how Claude, ChatGPT, or Cursor executes Read, Prepare, and Action tools against Future MCA.

35 Registered Tools
Level 1: Read Tools
Level 2: Prepare Tools
Level 3: Action Execution