Skip to content
0%

Alchemy plugin for Claude Code now live

Author: Lisa Ma

Last updated: June 22, 20264 min read
Alchemy Claude Plugin

Traditional developer flows are evolving. They used to look like this: you’re coding and you need something - a wallet balance, the current price of ETH, finding whether a transaction actually landed. So you stop. You open the dashboard, dig through docs, paste together an SDK call, switch chains, and eventually get your answer.

We built the Alchemy plugin for Claude Code to delete that round trip. Install it once, and your coding agent can query blockchain data across 100+ chains and manage your Alchemy apps without ever leaving the conversation. Ask "what's the current price of ETH?" or "show me the NFTs owned by vitalik.eth" and you get a real answer, from live data, right where you are already working.

The plugin is open source and available now at alchemyplatform/alchemy-claude-plugin.

What you get in one install

One install gives Claude Code three things at once. It connects the hosted Alchemy MCP Server, so Claude can call live blockchain data as tools. It adds slash commands for the queries you run most, so common work is one command away. And it ships Agent Skills that teach Claude how to use Alchemy correctly, both for answering questions in the conversation and for writing application code against our APIs.

You do not configure any of that. You install, sign in, and start asking.

Getting started

Inside Claude Code, add the marketplace and install the plugin:

bash
Copied
/plugin marketplace add alchemyplatform/alchemy-claude-plugin /plugin install alchemy@alchemy

The first time Claude calls an Alchemy tool, a browser window opens and you sign in with your Alchemy account over OAuth. No API key to copy, no local server to run. If you do not have an account yet, you can create one for free at the Alchemy Dashboard.

Then run setup once to pick which app you are working with:

/alchemy:setup

That caches the app's API key for the rest of your session, which is what most of the data tools need. After that you are free to ask in plain English or reach for a command.

The commands you will actually use

The slash commands cover the lookups that interrupt you most. Each one takes an optional network and defaults to Ethereum mainnet when you leave it off.

Command
What it does
/alchemy:setup
Connect and select an Alchemy app for the session
/alchemy:balance [network]
Native and ERC-20 balances with USD values
/alchemy:nfts [network]
NFTs owned by a wallet
/alchemy:token [network]
Token price and metadata
/alchemy:tx [network]
Transaction details and receipt
/alchemy:portfolio [networks]
Multi-chain tokens and NFTs
/alchemy:gas [network]
Current gas and fee estimates
/alchemy:solana
SOL balance and SPL token accounts
/alchemy:create-app
Create a new Alchemy app

Read across chains with portfolio. Check a balance in USD without wiring up a price feed. Look up a transaction and its receipt to confirm it settled. Drop into Solana for SOL and SPL token accounts in the same flow you use for EVM. And when you are ready to ship, create-app provisions a new Alchemy app without leaving the terminal.

Behind each command sits 160+ tools from the hosted Alchemy MCP Server: EVM JSON-RPC, our Token, NFT, Prices, and Portfolio APIs, transaction simulation and tracing, gasless transactions, and Solana over both RPC and the Digital Asset Standard. And, no need to memorize commands. Ask Claude directly and it will reach for the right tool.

Why this matters

The point is not that you can look up a balance. You always could. The point is that the lookup now happens inside the work instead of beside it.

When your agent can read the chain in context, the questions change. You stop asking "what is this address holding" as a separate errand and start asking it as part of debugging a failing flow. You can have Claude check a transaction receipt, reason about why it reverted, and propose a fix without you ever leaving the file. Simulation and tracing are right there, so you can understand what a transaction will do before you send it. The agent is not just writing code about the chain. It can see the chain.

That is also why we ship the Agent Skills alongside the tools. Live data answers questions in the moment. The skills teach Claude the patterns behind our APIs, so when it writes code against Alchemy, it writes the kind of code we would. You get correct help in the conversation and correct code in your repo.

If you do not use Claude Code

The plugin is the easiest path for Claude Code users, but it is one of three ways to work with Alchemy from an agent. Pick the one that matches where you build.

Best for
Claude plugin
Claude Code, where one install bundles the MCP server, slash commands, and skills
The terminal directly, for live queries, transactions, and admin tasks in shell scripts, cron jobs, and CI pipelines
Any MCP-compatible client such as Cursor, Codex, Claude Desktop, or VS Code Copilot, for live blockchain data as tool calls
Any agent that writes code against Alchemy APIs, for endpoint knowledge without tool calls

If you are in Claude Code, install the plugin and the MCP server, slash commands, and skills all come together.

Try it

Add the marketplace, install, run /alchemy:setup, and ask Claude something onchain. The whole thing takes about a minute, the account is free, and the plugin is open source if you want to see how it works or contribute.

Get it at alchemyplatform/alchemy-claude-plugin, or read the docs.

Alchemy Newsletter

Be the first to know about releases

Sign up for our newsletter

Get the latest product updates and resources from Alchemy

A
O
D
+
Over 80,000 subscribers

By entering your email address, you agree to receive our marketing communications and product updates. You acknowledge that Alchemy processes the information we receive in accordance with our Privacy Notice. You can unsubscribe anytime.