- SANDBOX
- DEBUG
- VERIFY
- HUMAN
Short answer
How can I use AI to make virtual testnets?
You tell your AI to replicate the Ethereum mainnet state at a specific block. It uses Tenderly to build a private, isolated sandbox where you can test dApps without using real funds. This setup lets you debug complex interactions in a controlled environment.
Testing outcomes
Where your testing lands.
What you get once the AI finishes the setup.
SANDBOX
Isolated environments
The AI builds a private replica of the blockchain. You get a clean space to run transactions without affecting the real network.
DEBUG
State inspection
Your agent accesses the specific block state requested. You can see exactly how your contract interacts with existing liquidity or protocols.
VERIFY
Transaction testing
The AI executes your code against the virtual state. You receive immediate feedback on whether your logic holds up against real-world data.
HUMAN
Manual review
The AI handles the heavy lifting of environment setup. You step in to inspect the Tenderly dashboard and finalize your contract audits.
The workflow
What your AI does when the request arrives.
The AI handles the technical setup of the Tenderly environment.
Define state
The AI identifies the specific Ethereum block height you need to replicate for your test.
create_virtual_testnetDeploy replica
Your agent instructs Tenderly to build the virtual testnet based on that block's state.
create_virtual_testnetConfigure environment
The AI prepares the sandbox for your specific dApp deployment and testing needs.
create_virtual_testnetReport status
Your agent confirms the virtual testnet is live and ready for your contract interactions.
create_virtual_testnet
Try it
Copy these to start.
Use these prompts to kick off your testing workflow.
Starting points
Swap out the block numbers and contract addresses to match your specific testing scenario.
Tenderly Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_N4CrFBeUVzbat4muj6uWeqLKjZIfSS9VAScUfQGd/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Tenderly capabilities are ready to use.
{
"mcpServers": {
"tenderly-ethereum-dev-platform-mcp": {
"url": "https://edge.vinkius.com/vk_preview_N4CrFBeUVzbat4muj6uWeqLKjZIfSS9VAScUfQGd/mcp"
}
}
}Create a virtual testnet at block 18000000 so I can test my swap logic.
Spin up a private mainnet replica using the latest block to test this contract address: 0x123...
I need an isolated environment at a specific block to debug a failed transaction.
Set up a virtual testnet for me so I can deploy my new dApp in a mainnet-like state.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Tenderly once, then ask.
Link your Tenderly account to your AI client in one step. Once connected, you can manage your testing environments through natural language.
Connect Tenderly to your AIFAQ
How this task behaves.
- 01
Can the AI deploy contracts directly to Ethereum mainnet?
No. The AI only uses Tenderly to create virtual, isolated testnets for safe experimentation.
- 02
Does the AI manage my Tenderly API keys?
No. You connect your account through Vinkius, and your credentials stay encrypted and managed by the platform.
- 03
Can I use this to test against a specific historical block?
Yes. You can instruct the AI to use a specific block number when creating the virtual testnet.
- 04
Will the AI delete my existing Tenderly projects?
No. The AI only has the capability to create new virtual testnets; it cannot modify or delete your existing Tenderly configurations.
- 05
Do I need to write code to trigger the testnet creation?
No. You just tell your AI what kind of environment you need, and it handles the tool execution.
More questions about Tenderly? The Connector page answers them. See everything the Tenderly Connector can do
Connect Tenderly to Claude, Cursor, ChatGPT & more
