Skip to content
plugmcp
DirectorySkillsCategoriesBlog
EN/中文
Sign inSign up
Home/Servers/MetaMesh UGA

MetaMesh UGA

by dev.metamesh-uga

Serverless edge control plane to discover, route and monetize 14,982+ MCP tools.

Power your agent

Add this server

The publisher's own URL. Not a PlugMCP endpoint.

Server URL
https://api.metamesh-uga.dev/mcp
Provider
Add in Claude

Claude opens with the name and URL filled in. You still confirm it there.

Claude Code
claude mcp add --transport http dev.metamesh-uga-metamesh-uga https://api.metamesh-uga.dev/mcp

Run in a terminal.

Claude Desktop
{
  "mcpServers": {
    "dev.metamesh-uga-metamesh-uga": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://api.metamesh-uga.dev/mcp"
      ]
    }
  }
}

claude_desktop_config.json

Step-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Related MCP

View more
  • Firecrawl🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
  • N
    NotionToken-auth Notion MCP: pages, databases, blocks, comments, files. Runs headless in CI and agents.
  • Context7Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors

Related skills

View all
  • reaReverse engineer anything with agents, from app behavior down to native binaries.
  • rea-changelog-updatePrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.
  • rea-tool-designDesign or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.

Questions

How do I install MetaMesh UGA?
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
Does MetaMesh UGA need an API key?
No required environment variable is listed on this record.
What transport does MetaMesh UGA use?
MetaMesh UGA is reached over HTTP.
When was MetaMesh UGA last checked?
Last checked 2026-10-16. Serverless edge control plane to discover, route and monetize 14,982+ MCP tools.

Serverless edge control plane to discover, route and monetize 14,982+ MCP tools.

MetaMesh-UGA

The MCP Operating System — A Serverless Control Plane for AI Agents and MCP Infrastructure

Set & Forget | Zero Infrastructure | Serverless | Edge Native | MCP Operating System

Live Status

  • Gateway: https://api.metamesh-uga.dev — operational
  • Dashboard: https://dashboard.metamesh-uga.dev
  • MCP endpoint: https://api.metamesh-uga.dev/mcp (JSON-RPC over HTTP)
  • MCP Registry: dev.metamesh-uga/metamesh-uga v2.0.0 — published & latest
  • Tools indexed: 13,000+ MCP servers, synced every 6h from the official MCP registry
  • Stripe billing: LIVE — agents top up a USD balance and are charged per call
  • x402 payments: implemented; activation gated by wallet config (see /v1/x402/info)

MetaMesh-UGA is the MCP Operating System: a serverless, edge-native control plane that discovers, verifies, scores, routes, monitors and scales MCP servers behind a single endpoint.

It combines registry sync, semantic discovery, trust scoring, policy enforcement, security scanning, lifecycle management, smart routing, multi-level caching, reliability patterns, self-healing and real-time analytics to make MCP infrastructure production-ready for AI agents.

Connect in seconds: point any MCP client (Claude, Windsurf, Cursor, Antigravity) at https://api.metamesh-uga.dev/mcp, or register an agent for a programmatic API key. Either way you gain access to a curated, trusted, and benchmarked catalog of MCP servers.

🚀 Caratteristiche

  • MCP Operating System: Control Plane + Data Plane per gestire l'intero ecosistema MCP
  • Unified Registry: Single source of truth con federation, mirroring e snapshot
  • Smart Discovery: Keyword, semantic search, capability graph e intent-based discovery
  • Trust Engine: Reputation score basato su uptime, latenza, success rate e security
  • Smart Routing: Weighted, latency, geographic, cost e health-based routing
  • Reliability Layer: Circuit breaker, retry, bulkhead, hedging e fallback
  • Multi-level Cache: In-memory L1 + KV L2, TTL, pattern invalidation
  • Security Fabric: Dependency scan, CVE detection, policy engine OPA/Rego
  • Self-Healing: Auto-deprecated, rollback recovery, health-based remediation
  • Analytics: Usage trends, cost analysis, performance metrics, export OTel/Prometheus
  • Lifecycle Management: Stati automatici dalla discovery all'archiviazione
  • Serverless: Zero infrastruttura da gestire (Cloudflare Workers)
  • WASM Runtime: Esecuzione sicura dei tool MCP compilati in WebAssembly
  • AI Agent Economy: Supporto nativo per x402 Protocol (micropagamenti USDC)
  • Monetizzazione: Piani Free, Pro ($19/mese), Enterprise ($499/mese)
  • Set & Forget: Automazione completa con cron triggers

📋 Requisiti

  • Node.js 18+
  • Go 1.21+ (per CLI)
  • Account Cloudflare (free tier sufficiente)
  • Account Stripe (per pagamenti)

🛠️ Installazione

1. Clone e Setup

git clone https://github.com/SiouxTXT/metamesh-uga.git
cd metamesh-uga
npm install

2. Configurazione Cloudflare

# Installa Wrangler
npm install -g wrangler

# Login a Cloudflare
wrangler login

# Crea database D1
wrangler d1 create metamesh-catalog

# Crea KV namespace
wrangler kv:namespace create "CACHE"

# Crea R2 bucket
wrangler r2 bucket create metamesh-wasm

3. Configura Secrets

# Stripe
wrangler secret put STRIPE_SECRET_KEY
wrangler secret put STRIPE_WEBHOOK_SECRET

# JWT
wrangler secret put JWT_SECRET

# Alerting (opzionale)
wrangler secret put TELEGRAM_BOT_TOKEN
wrangler secret put TELEGRAM_CHAT_ID

# Admin
wrangler secret put ADMIN_KEY

4. Database Migration

wrangler d1 execute metamesh-catalog --file=./shared/migrations/001_init.sql

5. Deploy

# Deploy Workers
npm run deploy:workers

# Deploy Dashboard
cd packages/dashboard
npm install
npm run deploy

6. Installa CLI

cd cli
go build -o metamesh

# Installa globalmente
mv metamesh /usr/local/bin/

# O usa direttamente
./metamesh connect

🎯 Uso

CLI Tool

# Connetti al gateway (salva la config locale)
metamesh connect

# Chiama un tool (raggiunge POST /v1/call)
metamesh call example.echo --message "hello"

# Lista tool disponibili
metamesh list --category demo

# Visualizza utilizzo
metamesh usage

API REST

# List tools (paginated, public)
curl "https://api.metamesh-uga.dev/v1/tools?limit=20&sort=popularity"

# Full-text search
curl "https://api.metamesh-uga.dev/v1/search?q=github"

# Execute a tool
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{"tool": "example.echo", "params": {"message": "hello"}}' \
  https://api.metamesh-uga.dev/v1/call

MCP Protocol

Add to Devin / Cursor / Windsurf / Claude Desktop:

{
  "mcpServers": {
    "metamesh-uga": {
      "url": "https://api.metamesh-uga.dev/mcp",
      "transport": "http"
    }
  }
}

Or test directly via JSON-RPC:

curl -X POST https://api.metamesh-uga.dev/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

🤖 AI Agent Economy

Two complementary, production-ready payment models:

Model A — Stripe prepaid credit (recommended)

Agents top up a USD balance once via Stripe Checkout, then per-call pricing is debited automatically from the balance. No per-call fees, no crypto required.

# 1. Register an agent (returns agent_id + api_key, shown once)
curl -X POST https://api.metamesh-uga.dev/v1/agent/register \
  -H "Content-Type: application/json" -d '{"name":"My Agent"}'

# 2. Top up via Stripe (returns a Checkout URL)
curl -X POST https://api.metamesh-uga.dev/v1/agent/topup \
  -H "X-Agent-Id: agent_xxx" -H "X-Agent-Key: ak_xxx" \
  -H "Content-Type: application/json" -d '{"amount_usd": 10}'

# 3. Call tools — balance is debited per call
curl -X POST https://api.metamesh-uga.dev/v1/call \
  -H "X-Agent-Id: agent_xxx" -H "X-Agent-Key: ak_xxx" \
  -H "Content-Type: application/json" \
  -d '{"tool":"example.echo","params":{"message":"hi"}}'

# 4. Check balance / history
curl https://api.metamesh-uga.dev/v1/agent/wallet -H "X-Agent-Id: agent_xxx" -H "X-Agent-Key: ak_xxx"
curl https://api.metamesh-uga.dev/v1/billing/info

Operator activation:

wrangler secret put STRIPE_SECRET_KEY --env production       # sk_live_... or sk_test_...
wrangler secret put STRIPE_WEBHOOK_SECRET --env production   # whsec_... (Stripe webhook signing secret)
# Point a Stripe webhook to: https://api.metamesh-uga.dev/v1/stripe/webhook (event: checkout.session.completed)
# Optionally set BILLING_ENABLED="true" in wrangler.toml to require payment for ALL callers.

When Stripe is unset, anonymous calls stay free and /v1/agent/topup returns a clear 503.

Current production state: Stripe (live keys) and the checkout.session.completed webhook are configured, and BILLING_ENABLED="true". Enforcement applies to the REST POST /v1/call endpoint — every caller is charged per request, so top up via /v1/agent/topup first. The public MCP endpoint (POST /mcp) stays free for discovery (tools/list) and native metamesh.* tools, so MCP clients keep working without an account.

Model B — x402 crypto (USDC on Base)

The x402 protocol: agents pay per call in USDC on-chain. Send an X-PAYMENT header; on HTTP 402 read the accepts requirements.

curl https://api.metamesh-uga.dev/v1/x402/info

Payment flow

  1. Call a paid resource (POST /v1/call). When enforcement is active and no payment is attached, the gateway responds HTTP 402 with a standard accepts requirements body (scheme exact, network base, asset USDC, payTo, maxAmountRequired).
  2. The agent constructs a signed payment and retries with an X-PAYMENT header (base64 JSON).
  3. The gateway verifies + settles via the configured facilitator, then returns the result plus an X-PAYMENT-RESPONSE header with the on-chain tx.

Activation (operator)

x402 is implemented but disabled until you set a receiving wallet:

# In wrangler.toml [env.production.vars]
X402_ENABLED = "true"
X402_PAY_TO  = "0xYourBaseWallet"
# X402_FACILITATOR_URL and USDC_CONTRACT have working defaults

Per-tool pricing lives in the tool_pricing table (default $0.001/call, bulk $0.0005).

📊 Architettura

MetaMesh-UGA adotta un'architettura a due piani ispirata ai service mesh moderni.

┌─────────────────────────────────────────────────────────────────────────────┐
│                      META MESH-UGA — MCP OPERATING SYSTEM                 │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│  ┌─────────────────────────────────────────────────────────────────────┐   │
│  │                     CONTROL PLANE                                   │   │
│  │                                                                     │   │
│  │  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │   │
│  │  │ Registry │ │ Discovery│ │  Trust   │ │  Policy  │ │ Lifecycle│ │   │
│  │  │  Engine  │ │  Engine  │ │  Engine  │ │  Engine  │ │  Manager │ │   │
│  │  └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ │   │
│  │  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐             │   │
│  │  │ Security │ │Benchmark │ │Analytics │ │  Config  │             │   │
│  │  │  Scanner │ │  Engine  │ │  Engine  │ │  Engine  │             │   │
│  │  └──────────┘ └──────────┘ └──────────┘ └──────────┘             │   │
│  └─────────────────────────────────────────────────────────────────────┘   │
│                                    │                                        │
│                                    ▼                                        │
│  ┌─────────────────────────────────────────────────────────────────────┐   │
│  │                       DATA PLANE                                    │   │
│  │                                                                     │   │
│  │  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │   │
│  │  │  Routing │ │   Cache  │ │  Health  │ │  Proxy   │ │  Rate    │ │   │
│  │  │  Engine  │ │  Engine  │ │  Engine  │ │  Engine  │ │ Limiting │ │   │
│  │  └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ │   │
│  └─────────────────────────────────────────────────────────────────────┘   │
│                                    │                                        │
│                                    ▼                                        │
│  ┌─────────────────────────────────────────────────────────────────────┐   │
│  │                         MCP ECOSYSTEM                               │   │
│  │  (Server MCP A, B, C, D, E...)                                     │   │
│  └─────────────────────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────────────────────┘
  • Control Plane: decisioni, configurazione, orchestrazione, analytics.
  • Data Plane: esecuzione delle richieste, routing, caching, health in tempo reale.

Per maggiori dettagli: docs/ARCHITECTURE.md.

💰 Piani di Pagamento

PianoPrezzoFeature
Free€01.000 chiamate/mese, 100 req/min
Pro$19/meseChiamate illimitate, 1.000 req/min
Enterprise$499/meseDedicated VPC, 10.000 req/min, SLA 99.9%

📁 Struttura Repository

metamesh-uga/
├── packages/
│   ├── gateway/            # Data Plane: API Gateway, Routing, Cache, Proxy, Reliability
│   ├── discovery/          # Control Plane: Discovery Engine
│   ├── registry/           # Control Plane: Registry Engine (federation, mirroring, snapshot)
│   ├── trust/              # Control Plane: Trust Engine
│   ├── security/           # Control Plane: Security Scanner
│   ├── policy/             # Control Plane: Policy Engine
│   ├── benchmark/          # Control Plane: Benchmark Engine
│   ├── lifecycle/          # Control Plane: Lifecycle Manager
│   ├── recommendation/     # Control Plane: Recommendation Engine
│   ├── cost/               # Control Plane/Data Plane: Cost Optimizer
│   ├── compatibility/      # Control Plane: Compatibility Engine
│   ├── intelligence/       # Control Plane: AI Intelligence Layer
│   ├── analytics/          # Control Plane: Analytics Engine
│   ├── health/             # Data Plane: Health Engine
│   ├── config/             # Control Plane: Config Engine
│   ├── cache/              # Data Plane: Advanced Cache Engine
│   ├── self-healing/       # Data Plane: Self-Healing Engine
│   ├── aggregator/         # Motore di aggregazione
│   ├── inserter/           # Motore di inserimento
│   ├── updater/            # Motore di aggiornamento
│   ├── eliminatore/        # Motore di eliminazione
│   ├── alerts/             # Alerting Worker
│   ├── agent-billing/      # Billing per AI Agent
│   ├── dashboard/            # Cloudflare Pages (React)
│   └── landing/            # Landing page
├── cli/                    # CLI Tool (Go)
├── shared/                 # Librerie condivise
│   ├── src/
│   │   ├── auth.js         # Autenticazione
│   │   ├── ratelimit.js    # Rate limiting
│   │   ├── wasm-runtime.js # WASM execution
│   │   └── x402.js         # x402 Protocol
│   └── migrations/         # Database migrations
├── docs/                   # Documentazione
└── wrangler.toml           # Configurazione Cloudflare

🔧 Configurazione Environment

# .env
CLOUDFLARE_API_TOKEN=xxx
CLOUDFLARE_ACCOUNT_ID=xxx
D1_DATABASE_ID=xxx
JWT_SECRET=xxx
STRIPE_SECRET_KEY=sk_test_xxx
STRIPE_WEBHOOK_SECRET=whsec_xxx
TELEGRAM_BOT_TOKEN=xxx
TELEGRAM_CHAT_ID=xxx

🧪 Testing

# Test locale
wrangler dev --config packages/gateway/wrangler.toml

# Test API
curl http://localhost:8787/health

📝 Licenza

MIT License - vedi LICENSE

👤 Autore

Davanzo Keoma - GitHub


Il "set & forget" è realtà. 🚀

How do I install MetaMesh UGA?
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
Does MetaMesh UGA need an API key?
No required environment variable is listed on this record.
What transport does MetaMesh UGA use?
MetaMesh UGA is reached over HTTP.
When was MetaMesh UGA last checked?
Last checked 2026-10-16. Serverless edge control plane to discover, route and monetize 14,982+ MCP tools.

No tool list yet. Remote servers get one after the first health check.

plugmcp

Plug any tool into any agent. MCP servers for people who do not write servers. Gateway for volume, n8n for apps.

  • github.com/TODO
  • x.com/TODO
  • linkedin.com/company/TODO

PlugMCP, 2026

Product
  • Browse
  • Install
  • Host your own
  • Pricing
  • Voice mode
  • Browser
Directory
  • All MCP servers
  • MCP Apps
  • Tools
  • Agent skills
  • Categories
  • New
  • Voice
  • Submit a server
Developers
  • Docs
  • Blog
  • Club
  • Work with us
  • What is MCP
  • What is WebMCP
  • Status
Company
  • About
  • Contact
  • Terms
  • Privacy
  • DPA