blockrun
BlockRun MCP Server - Access 30+ AI models in Claude Code with zero API keys. Pay-per-request via x402 micropayments.
claude mcp add --transport stdio blockrunai-blockrun-mcp npx -y @blockrun/mcp
How to use
BlockRun MCP provides access to 30+ AI models across multiple providers without separate API keys or multiple accounts. You install the MCP package once, and then issue a single command to load BlockRun into Claude Code (via the claude mcp workflow). Once running, you can chat with OpenAI, Claude, Gemini, DeepSeek, and other models, generate images, list available models, monitor budgets, and manage a single wallet for pay-per-use transactions on Base. The tooling includes smart model selection, real-time web search capabilities via Grok, wallet balance checks, and budget controls to cap spending per session. The commands are designed to be intuitive: you can ask a model directly, start a chat, generate an image, or have BlockRun pick the best model for a given task, such as fast responses, daily tasks, or complex analysis. Use the wallet and balance commands to ensure you have funds, and fund your wallet on the Base network as needed.
How to install
Prerequisites:
- Node.js and npm (for npx-based installation) or a Claude Code environment capable of running MCP commands
- Access to Claude CLI (claude mcp) if you are integrating through Claude
- Optional: an existing wallet key (BASE_CHAIN_WALLET_KEY or BLOCKRUN_WALLET_KEY) if you want to use your own wallet
Step-by-step:
- Ensure prerequisites are installed:
- Install Node.js: https://nodejs.org/
- Confirm npm is available: npm -v
- Install Claude CLI if you haven’t already and sign in to Claude.
- Add the BlockRun MCP server: claude mcp add blockrun npx @blockrun/mcp This installs the MCP and wires it into Claude’s MCP workflow.
- (Optional) Provide a wallet key if you want to use an existing wallet: claude mcp add blockrun npx @blockrun/mcp --env BLOCKRUN_WALLET_KEY=0x...
- Initialize a wallet (the first-time setup will create a wallet if none exists): blockrun setup
- Fund your wallet on the Base network following the on-screen instructions and the wallet address provided by blockrun setup.
- Start using BlockRun via the available commands, e.g.:
- blockrun ask GPT-5 to explain quantum computing
- blockrun list models
- blockrun generate an image of a cyberpunk cityscape
Additional notes
Notes and tips:
- Wallet funding: You can start with as little as $5 worth of USDC on Base. Monitor your spend with blockrun budget commands.
- No API keys are required for supported models; all interactions are billed to a single wallet.
- If you encounter payment or wallet issues, ensure the wallet file (~/.blockrun/.session) is writable or re-run blockrun setup to regenerate.
- The wallet priority is: BLOCKRUN_WALLET_KEY environment variable, then BASE_CHAIN_WALLET_KEY, then the local session file, then auto-generation.
- You can customize the wallet by exporting BLOCKRUN_WALLET_KEY or BASE_CHAIN_WALLET_KEY or by passing them via claude mcp add with --env. Verify transactions on Basescan.
- Smart model selection can drastically reduce cost and latency by routing requests to the most suitable model (fast, balanced, powerful, cheap, or reasoning).
- The MCP supports chat, image generation, model listing, and real-time web search (T/X) integration.
Related MCP Servers
time
⏰ Time MCP Server: Giving LLMs Time Awareness Capabilities
mcp-tasks
A comprehensive and efficient MCP server for task management with multi-format support (Markdown, JSON, YAML)
lichess
A Model Context Protocol (MCP) server that enables Claude Desktop to interact with Lichess chess platform through natural language. Play games, analyze positions, manage your account, and participate in tournaments—all by simply talking to Claude.
HydraMCP
Connect agents to agents. MCP server for querying any LLM through your existing subscriptions: compare, vote, and synthesize across GPT, Gemini, Claude, and local models from one terminal.
claude-code-agentic-semantic-memory-system
This guide provides complete instructions for implementing an **Agentic Semantic Memory System** that enables Claude agents to:
AI-web mode
一个基于 MCP (Model Context Protocol) 的智能对话助手Web应用,支持实时聊天、工具调用和对话历史管理。