agvps

AI 에이전트용 서버, 월 단위 임대: 루트 액세스 권한과 공용 IPv4를 갖춘 Linux 서버. 계정 없이, x402 스테이블코인으로 호출당 결제하며, 다시 결제하면 갱신됩니다.

호스팅형 MCP 서버

npx add-mcp 'https://agvps.shveik.dev/mcp'

Claude Code, Codex, Cursor 등에 설치됩니다

문서

Linux servers for agents, rented by the month.

Pay in stablecoins with x402 or MPP and get a server with root access and a public IPv4 address. Pay again to keep it.

1. Ask

POST the plan and the location. Without a payment you get the quote.

2. Pay

Sign the quote with x402 or MPP and send the same request again.

3. Receive

The server is ordered only after the payment settles. It is ready in 5 to 10 minutes.

4. Renew

Pay again for 1 to 6 more months. A call that cannot be served is not charged.

Good to know

  • Payment: Stablecoins
  • Accounts: None needed
  • Protocols: x402 and MPP
  • Run by: shveik.dev
  • Compared with: Server rental services generally organize orders around accounts; agvps quotes each order for stablecoin payment and orders after settlement.

Authentication/payment

No API key or account is required. The unpaid request returns HTTP 402 with a quote: x402 uses PAYMENT-REQUIRED and a paid retry with PAYMENT-SIGNATURE; MPP uses WWW-Authenticate: Payment and a paid retry with Authorization: Payment. Review and sign the quote, then retry the same request. See the deprecation policy.

Pricing

Current endpoint prices and limits are generated at https://agvps.shveik.dev/pricing.md. Each quote is returned before payment.

Endpoints

  • POST /x402/vps-small (x402) and POST /mpp/vps-small (MPP): Server, 1 vCPU 1 GB RAM
  • POST /x402/vps-small-renew (x402) and POST /mpp/vps-small-renew (MPP): Renew a small server
  • POST /x402/vps-medium (x402) and POST /mpp/vps-medium (MPP): Server, 2 vCPU 2 GB RAM
  • POST /x402/vps-medium-renew (x402) and POST /mpp/vps-medium-renew (MPP): Renew a medium server

OpenAPI

OpenAPI 3.1

MCP

Remote MCP is available at https://agvps.shveik.dev/mcp, with tools corresponding to the enabled endpoints.

Rate limits

Free documents are limited to 120 requests per IP per minute. Paid API routes have no general request quota; endpoint-specific limits are documented with each route.

Limits and reliability

No uptime SLA; service and upstream availability are best-effort. Free GET documents, including /docs, /agents.md, /llms.txt, /openapi.json, /pricing and /pricing.md, are limited to 120 requests per client IP per minute and return RateLimit headers; HTTP 429 includes Retry-After. Paid-route timeouts and any route-specific in-flight limits are described in the product docs. GET /health is a liveness check on agdata, agproxy and agvps; agmail and agpay use GET /healthz. HTTP 402 means payment is required, 429 means a limit was reached, 502 means delivery failed, and 503 means the service cannot accept work; follow Retry-After when present. Review the quote before paying. An unpaid request returns a quote before payment; work starts after settlement. A request refused before settlement is not charged; product-specific refund rules apply.

Capabilities

Linux servers for AI agents, rented by the month. Pay in stablecoins over x402 or MPP and get root access and a public IPv4 address.

Contact

Contact the operator. Read the full documentation