Wingman MCP Server: A New Paradigm for Pay-Per-Use AI Services Based on the x402 Protocol

Wingman MCP Server is an open-source project developed by RIPRODUCTIONS, which implements account-free, subscription-free on-demand pay-per-call AI services via the x402 payment protocol on the Base L2 network. It covers functions like image generation, LLM dialogue, web scraping, SEO audit, and content generation, using USDC as the payment currency. This project represents an important exploration of AI service payment models.

Key highlights: No account registration or payment binding required; AI agents can automatically complete payments through x402; leverages Base L2's low transaction fees and fast confirmation speed.

With the popularization of AI applications, traditional subscription-based payment models have gradually exposed flexibility issues. Many developers and small teams do not need fixed monthly API call quotas but hope to pay based on actual usage, leading to the rise of pay-per-call models.

However, achieving true on-demand payment is not easy. Traditional API key methods require users to register accounts and bind payment methods, which is cumbersome. Cryptocurrency and blockchain technology provide new ideas—realizing account-free instant payments through smart contracts and payment protocols.

Wingman MCP Server is built on the x402 payment protocol, with the core concept of enabling AI agents to automatically complete payments via x402 without accounts or subscriptions. It is deployed on the Base L2 network and uses USDC as the payment currency.

Current services and pricing:

• Image Generation: $0.10 per call (uses Flux Schnell model via ComfyUI)
• LLM Chat: $0.002 per call (based on Ollama-run llama3.1 model)
• Web Scraping: $0.03 per call (supports structured data extraction from any URL)
• SEO Audit: $0.05 per call (provides technical SEO analysis and 0-100 score)
• Content Generation: $0.01 per call (supports blog posts, social media copy, etc.)

The x402 protocol is the core of Wingman MCP Server, defining how to embed payment information in HTTP requests so the server can verify payment status before responding.

The project exposes service discovery and pricing information via the /.well-known/x402 endpoint, allowing clients to get supported services, prices, and receiving wallet addresses. Payments are settled using USDC on Base L2 (low Gas fees, fast confirmation), with facilitators from x402.org handling payment verification and settlement.

Wingman MCP Server uses a Python tech stack: FastAPI framework with x402 middleware for payment verification (high-performance asynchronous handling). It integrates components like Ollama (local LLM operation) and ComfyUI (image generation with Flux Schnell).

Data storage: SQLite for request logs (zero configuration, easy migration to PostgreSQL for higher performance). Deployment steps: Install dependencies (fastapi, uvicorn, httpx, x402, beautifulsoup4, stripe) and start via uvicorn.

Wingman MCP Server is suitable for:

1. Autonomous payment needs of AI agents (no human intervention required)
2. On-demand payment in microservice architectures (precise billing for inter-service calls)
3. Developer tools and automation workflows (more cost-effective than fixed subscriptions for frequent AI API calls)

The project is open-source under the MIT license, allowing free use, modification, and distribution. It provides a complete reference implementation for developers who want to build their own AI service infrastructure, combining blockchain payments with traditional Web services and facilitating community contributions.

Wingman MCP Server represents an important exploration direction for AI service payment models. Through the x402 protocol and Base L2 network, it achieves true account-free on-demand payment, opening new possibilities for AI service commercialization.

As crypto payment infrastructure matures and AI agent technology develops, similar pay-per-call models may become industry standards. Developers and enterprises are advised to understand and try this new model to gain an advantage in the future AI economy.