MindData Technology · Maestro Platform

Governed MCP Gateway
for Enterprise AI

MCProxy enforces authentication, authorization, quota, PII masking, credential injection, and full audit trail on every AI tool call.

Open Dashboard
POST /mcp — JSON-RPC 2.0
12
Modules
8
Pipeline Steps
MCP
Streamable HTTP
8-Step Request Pipeline

Every tool call, governed end-to-end

From authentication to audit, every step is enforced before reaching your upstream MCP servers.

01
AuthN
JWT / JWKS
02
AuthZ
Scopes / Roles
03
Quota
Rate limiting
04
Input Guard
PII masking
05
Cred Inject
Vault secrets
06
Route
Upstream MCP
07
Output Guard
Response filter
08
Trace
Full audit log
Platform Modules

Built for enterprise scale

🔐
M1 · Authentication
JWT validation with HS256 or remote JWKS. Supports multi-tenant identity extraction from token claims.
✅
M2 · Authorization
Tool-level scope enforcement. Per-server risk levels (low / medium / high) with approval workflows.
⏱
M3 · Quota & Rate Limit
Redis-backed sliding window rate limiting per user, tenant, or agent. Configurable per server.
🛡
M4 · Input Guard
PII detection and masking before arguments reach upstream servers. Credit cards, Thai IDs, emails and more.
🔑
M5 · Credential Injection
Vault-backed secret injection into tool calls. Agents never touch raw API keys or database credentials.
📊
M8 · Admin Portal
Real-time dashboard with server health, tool catalog, live audit log stream, and call metrics.

Ready to govern your AI?

Open the admin portal to monitor your MCP servers, inspect tool calls, and configure policies.

Open Admin Portal Check Gateway Health