Skip to main content
Glama

🏦 BankMCP: Secure Model Context Protocol Banking Gateway

DOI

License: BSL 1.1 Organization: RMAA Holdings Framework: FastMCP Security: DMCPS Hardened

BankMCP (engineered and maintained by RMAA Holdings) is an ultra-secure, Model Context Protocol (MCP) banking gateway built on TypeScript and FastMCP.

It enables AI Agents (Claude Desktop, ChatGPT, Gemini, ForgeAI) to query account balances, audit corporate ledgers, and prepare banking settlements without the catastrophic risk of unauthorized fund drainage or prompt injection attacks.


πŸ›‘οΈ DMCPS-Grade Security Rails

Bank APIs are high-consequence tools for autonomous agents. BankMCP adapts the hardened sandbox and dashboard interception architecture pioneered in DMCPS:

+---------------------------------------------------------------------------------------+
|                                BANKMCP INTERCEPTOR ARCHITECTURE                       |
+---------------------------------------------------------------------------------------+
|   AI AGENT (Claude / GPT / Gemini)                                                    |
|         β”‚                                                                             |
|         β–Ό  JSON-RPC (FastMCP / Stdio / Stream)                                        |
|   [BANKMCP SECURITY INTERCEPTOR] ──► Emergency Freeze Active? ──► [INSTANT BLOCK]     |
|         β”‚                                                                             |
|         β”œβ”€β”€β–Ί Non-Whitelisted IBAN? ─────────────────────────────► [INSTANT BLOCK]     |
|         β”œβ”€β”€β–Ί Exceeds Transaction Cap ($500)? ───────────────────► [INSTANT BLOCK]     |
|         β”œβ”€β”€β–Ί Exceeds Daily Cumulative Cap ($2,000)? ────────────► [INSTANT BLOCK]     |
|         β”‚                                                                             |
|         β”œβ”€β”€β–Ί Amount >= $100 (Human Gate)? ──────────────────────► [REQUIRES TOKEN]    |
|         β”‚                                                             β”‚               |
|         β”‚                                                             β–Ό               |
|         β”‚                                               [DMCPS SECURED DASHBOARD]     |
|         β”‚                                               Human Owner Approves Token    |
|         β”‚                                                             β”‚               |
|         β–Ό                                                             β”‚               |
|   [CORE BANKING / LEDGER EXECUTION] β—„β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜               |
+---------------------------------------------------------------------------------------+

1. Zero Direct Fund Drainage (Mathematical Whitelisting)

  • The agent cannot transfer funds to arbitrary accounts or IBANs found in prompts.

  • All recipients must pre-exist in the authorized Whitelist database, configured exclusively by a human administrator through the Web Dashboard.

2. Deterministic Velocity & Spending Caps

  • Single Transaction Cap: Default $500.00 maximum. Any single request above this is rejected at the schema level.

  • 24-Hour Rolling Quota: Cumulative daily transfers cannot exceed $2,000.00.

3. Human-In-The-Loop (HITL) Challenge Tokens

  • Any transfer equal to or greater than $100.00 generates a cryptographically signed challenge token (e.g. AUTH_9B2A...) with a 15-minute expiration.

  • The transfer is held in a pending state until a human logs into the Dashboard to approve it, or provides the challenge token to the submit_approval_token tool.

4. Emergency Killswitch (DEFCON-1 Freeze)

  • One-click freeze toggle on the dashboard immediately terminates all outbound transfer capabilities across all connected bank accounts.


Related MCP server: varco_mcp

πŸ› οΈ MCP Tools Exposed to AI

Tool Name

Risk Level

Description

list_accounts

🟒 Safe (Read)

Lists authorized bank accounts with balances and masked IBANs.

get_account_balance

🟒 Safe (Read)

Real-time balance check for a specific corporate account ID.

list_whitelisted_beneficiaries

🟒 Safe (Read)

Lists pre-approved recipient IBANs and their daily limits.

transfer_funds

πŸ”΄ High (Gated)

Drafts/executes transfer through whitelist, caps, and HITL gate.

submit_approval_token

🟑 Auth (Commit)

Submits human challenge token to finalize high-value transfer.

get_security_status

🟒 Safe (Read)

Audits active daily spent quotas, limits, and lockdown status.


πŸš€ Quickstart & Dashboard

1. Install Dependencies & Build

npm install
npm run build

2. Configure Environment

cp .env.example .env

Edit .env to set your administrator credentials:

PORT=8080
ADMIN_USERNAME=admin
ADMIN_PASSWORD=your_secure_password_here
MAX_TRANSACTION_LIMIT=500
MAX_DAILY_LIMIT=2000
REQUIRE_HUMAN_APPROVAL_ABOVE=100

3. Run the Security Gateway

npm start

The server will boot two coordinated services:

  • Admin Control Dashboard: http://localhost:8080/dashboard

  • FastMCP Transport Gateway: Stdio or Stream mode ready for MCP clients.

4. Run the Security Test Suite

npm test
βœ” BankSecurityGuard: Blocks non-whitelisted IBANs (16.9ms)
βœ” BankSecurityGuard: Enforces Single Transfer Cap (7.0ms)
βœ” BankSecurityGuard: Demands Human-In-The-Loop Approval Above Threshold (13.1ms)
βœ” BankSecurityGuard: Emergency Freeze Kills All Transfers Instantly (7.4ms)
β„Ή tests 4 | pass 4 | fail 0

πŸ“œ License

Released under the MIT License by RMAA Holdings & Contributors.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to analyze Ethereum wallets, simulate transactions, and draft transfers with deterministic policy and risk scoring, requiring human approval before on-chain execution.
      12 npm
      ISC
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables MCP-compatible AI agents to safely act on business backends by enforcing per-agent permissions, autonomy thresholds, human approval with review-and-edit, and full audit trails.
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents to propose structured payment intents, apply deterministic policy, require human approval, and record tamper-evident audit trails without connecting to live payment rails.
      MIT
    • A
      license
      A
      quality
      C
      maintenance
      Enables AI agents to query multichain balances, inspect smart contracts, and draft DeFi interactions while enforcing anti-drainer protections such as approval blocking, destination whitelisting, spend caps, and human approval gates.
      5
      Business Source 1.1