From Prompt to Transaction: Agentic Blockchain Workflows with EVMcrispr
Workshop: Simulate and Execute EVM Transactions with AI
Tired of manually assembling contract calls, encoding parameters, and praying your transaction doesn't fail on-chain? What if you could just describe what you want to do and have an AI agent write, validate, and simulate the entire transaction sequence before you sign anything?
In this hands-on workshop we'll explore EVMcrispr, a domain-specific language for encoding batched EVM transactions, and its new MCP (Model Context Protocol) server, which lets AI assistants like Claude interact directly with 60+ EVM-compatible blockchains.
What you'll learn
Write EVML scripts. EVML is a readable, human-friendly language for multi-step contract interactions, from simple token transfers to complex DAO governance proposals.
Simulate before you send. Fork any chain locally and test your transaction sequence in a sandboxed environment. Catch errors before they cost gas.
Let an AI agent do the heavy lifting. Connect the EVMcrispr MCP server to your AI assistant and go from natural language intent to validated, simulated, ready-to-sign transactions.
Execute on-chain. Once you're confident the simulation passes, sign and broadcast with your wallet.
What we'll cover
Introduction to EVML and the EVMcrispr terminal
Setting up the MCP server for agentic workflows
Writing and debugging scripts with AI assistance
Simulating forked chain transactions (EthereumJS backend)
Batching multiple contract calls into single transactions
Interacting with Aragon DAOs, ENS, and DeFi protocols
Going from simulation to real on-chain execution
Who is this for?
Developers, DAO contributors, and blockchain power users who want to move faster and safer when interacting with smart contracts. No prior EVMcrispr experience needed, just bring your laptop and a curiosity for what happens when AI meets on-chain execution.
Prerequisites
A browser wallet (MetaMask, Rabby, etc.)
An AI assistant that supports MCP (Claude Code, Claude Desktop, Cursor, VS Code + Copilot, etc.)
Food & Drinks
We'll have beer and pizza during the workshop so you can grab a slice, have a drink, and chat with other builders while experimenting with EVML, simulations, and AI-assisted transactions.
Links
MCP Server & AI Integration: mcp.evmcrispr.com
Web Terminal (preview): next.evmcrispr.com