Open Source CLI
Build Onchain AI Agents
One command to scaffold a full-stack monorepo with smart contracts, a frontend, and agent infrastructure.
$ npx scaffold-agent@latestFeatures
Smart Contracts
Foundry or Hardhat with auto-deploy, ABI generation, and type-safe contract hooks via Scaffold UI.
Frontend Ready
Next.js, Vite, or Python A2A with WalletConnect, Scaffold UI, shared Header nav, and suggested prompts.
9 LLM Providers
1Claw Shroud with 9 upstreams (OpenAI, Anthropic, Gemini, Mistral, Cohere, OpenRouter, Darkbloom, Venice) plus direct provider SDKs.
Multi-Chain Signing
1Claw Intents API with HSM/TEE signing across 29+ EVM chains plus Bitcoin, Solana, XRP, Cardano, and Tron.
Agent On-Chain Tools
Wallet balances, ENS resolution, ERC-8004 lookup, contract reads, 1Claw Intents, and optional Graph subgraph queries via Vercel AI SDK tools.
The Graph Integration
Optional subgraph data via --graph: MCP for IDE, x402 agent tools at runtime, /data page, and graph_search_subgraphs + graph_subgraph_query.
How It Works
- 1
Scaffold
$ npx scaffold-agent@latest my-agentThe interactive wizard walks through project name, secrets, agent identity, Ampersend, The Graph, LLM provider, chain framework, and app framework. Or use -y for defaults.
- 2
Quickstart
$ just quickstartOne command to start the local chain (background), fund wallets, deploy contracts, and launch the app. Or run just chain / just fund / just deploy / just start individually.
- 3
Validate
$ just doctorHealth check your environment — validates .env, tools, 1Claw IDs, package installs, and network config. Run anytime to diagnose issues.
- 4
Build
$ just startYour AI agent is live at localhost:3000 with chat UI, contract interaction, and on-chain tools. Modify agent logic in packages/ and iterate.
Point Your AI Here
Give your AI agent full knowledge of scaffold-agent by having it fetch the skill file. Add this to your project rules or tell your assistant directly:
Fetch https://scaffoldagent.xyz/SKILL.mdFrequently Asked Questions
- What is scaffold-agent?
- scaffold-agent is an open-source interactive CLI that scaffolds full-stack monorepo projects for building onchain AI agents. It generates smart contracts (Foundry or Hardhat), a frontend (Next.js, Vite, or Python), LLM integration, 1Claw MCP tools, and agent infrastructure in one command.
- How do I install it?
- No global installation required — just run npx scaffold-agent@latest in your terminal. You can also pass a project name directly: npx scaffold-agent@latest my-agent. The only prerequisites are Node.js, npm, and optionally the just task runner.
- What LLM providers are supported?
- 1Claw Shroud supports 9 upstream providers: OpenAI, Anthropic, Google/Gemini, Mistral, Cohere, OpenRouter, Darkbloom (E2E encrypted Apple Silicon TEE), and Venice (zero-retention + optional TEE). You can also use OpenAI, Anthropic, or Gemini directly. All chat routes use the Vercel AI SDK for streaming.
- What chains does the Intents API support?
- 1Claw Intents supports HSM/TEE transaction signing across 29 EVM mainnets (Ethereum, Base, Optimism, Arbitrum, Polygon, Avalanche, BNB, zkSync, Linea, Scroll, and more), EVM testnets, plus non-EVM chains: Bitcoin, Solana, XRP, Cardano, and Tron. Per-agent guardrails include chain restrictions, recipient allowlists, and daily spending limits.
- What is The Graph integration?
- Use --graph none|mcp|x402|both during scaffolding. MCP adds @graphprotocol/subgraph-mcp to your IDE config. x402 adds runtime agent tools (graph_search_subgraphs, graph_subgraph_query), a /data page, and lib/graph-client.ts. Optional GRAPH_API_KEY or vault path api-keys/thegraph; Substreams subgraphs may require x402 USDC per query.
- What is @1claw/mcp?
- When 1Claw is selected, scaffold-agent auto-generates .cursor/mcp.json and .mcp.json with the @1claw/mcp server pre-configured. This gives AI agents access to 44 MCP tools: vault secrets, Intents API (simulate, submit, sign), signing key management, treasury proposals, and execution intents.
- How do I point my AI assistant at scaffold-agent?
- Have your AI fetch https://scaffoldagent.xyz/SKILL.md — it contains the full CLI reference, all flags, project structure, just commands, LLM providers, Intents API, MCP integration, and agent tools. Add it to your project rules or tell your assistant directly.