Agent-Native Blockchain Infrastructure

These tools operate at different layers and are designed to be used together. Use x402 or MPP when your agent needs to read or write on-chain data without an account or API key. Use MCP when an assistant needs to create endpoints, check billing, or configure security. Use the Build Web3 plugin when an agent is generating code that calls Quicknode APIs. Use the Quicknode CLI when an agent works from a terminal and needs structured output it can parse. Use LLMs.txt when an LLM needs to find the right Quicknode documentation automatically. Read the Build with AI guide for a full comparison.

x402 is a wallet-authenticated, pay-per-request access layer built on the HTTP 402 standard. Any agent holding a wallet with USDC or USDG can start making RPC calls across 80+ chains immediately — no account registration, no API key. Payments are handled automatically by the @QuickNode Marketplace/x402 client: 1,000,000 credits for $10 USDC on mainnet. See the x402 payments guide for setup.

Two steps: install the Build Web3 plugin to give your coding agent accurate API knowledge (/plugin marketplace add quicknode/agent-plugins, then /plugin install build-web3@quicknode-agent-plugins in Claude Code, or codex plugin marketplace add https://github.com/quicknode/agent-plugins.git then codex plugin add build-web3@quicknode-agent-plugins in Codex), then access chains via x402 or a standard endpoint. On any other agent, install the skill alone with npx skills add quicknode/agent-plugins --skill build-web3. Read the Build Web3 plugin guide for details.

Quicknode's on-chain agent tools support many cases including: Autonomous Trading (real-time market data and execution across DEXs), Smart Contract Analysis (vulnerability and gas analysis), Conversational dApps (natural language blockchain interfaces), On-Chain Analytics (pattern detection and trend analysis via Streams), Wallet Intelligence (asset management and fee optimization), and Threat Monitoring (AI-powered anomaly detection on contracts and wallets). Each use case maps to a specific combination of Quicknode products. See the Build with AI guide for reference architectures.

The Build Web3 plugin bundles the build-web3 skill with the Quicknode MCP server, giving AI coding agents accurate, current knowledge of Quicknode's full product suite — so agents generate correct API calls on the first attempt rather than hallucinating method names or parameters. In Claude Code, install it with two commands: /plugin marketplace add quicknode/agent-plugins, then /plugin install build-web3@quicknode-agent-plugins. In Codex, the same plugin installs with codex plugin marketplace add https://github.com/quicknode/agent-plugins.git, then codex plugin add build-web3@quicknode-agent-plugins. On any other agent, install the skill on its own with npx skills add quicknode/agent-plugins --skill build-web3, no account required. The skill loads automatically at session start and works with Claude Code, Cursor, Codex, OpenCode, and Windsurf. Read the Build Web3 plugin guide for details.

The Quicknode MCP server (@QuickNode Marketplace/mcp) connects MCP-compatible AI assistants to your Quicknode account, letting them create and manage endpoints, configure security, set rate limits, and review billing through natural language — no dashboard required. It wraps the Admin API and is compatible with Claude Code, Claude Desktop, Cursor, and OpenAI Codex. It does not provide blockchain data access — for on-chain reads and writes, use x402 or a standard RPC endpoint. See the Quicknode MCP docs for setup steps.

Use HyperCore — Quicknode's data access layer for Hyperliquid L1. It provides four access methods: Info API (/info) for market data and positions, JSON-RPC (/hypercore) for block queries, WebSocket (/hypercore/ws) for real-time subscriptions, and gRPC (port 10000) for lowest-latency streaming. gRPC stream types include TRADES, ORDERS, BOOK_UPDATES, TWAP, EVENTS, BLOCKS, and WRITER_ACTIONS. Requires a Build plan or higher. See the Hyperliquid docs for method details.

Yes — five options depending on chain and latency requirements. Use Streams for deterministic event pipelines with guaranteed delivery. Use Yellowstone gRPC for sub-second Solana account and transaction monitoring. Use WebSocket connections for real-time subscriptions on EVM chains and Hyperliquid via HyperCore (/hypercore/ws). Use HyperCore gRPC for Hyperliquid order book and trade data. Use Webhooks for simple event-driven HTTP notifications.

Quicknode publishes machine-readable llms.txt indexes across its full documentation surface — products, chains, RPC methods, guides, sample apps, and courses — so AI systems reference the correct page rather than hallucinating or citing stale sources. Seven indexes are available, including the main platform index, the API reference index, and the x402 protocol index. Any docs or guides page can also be fetched as clean Markdown by appending .md to the URL. Read the LLMs.txt guide for the full list.