AI Tools

Remix has its own AI tool named RemixAI Assistant, which appears in the Right Side Panel once you log into your Remix account.

Access to RemixAI varies by your Remix plan:

• Free Plan: This provides access to Basic AI Skills and the QuickDApp Front-end Generator (excluding hosting), with AI model consumption billed on usage-based pricing.

• Starter & Pro Plans: Subscribing to our Starter or Pro plans unlocks more AI capabilities and includes an initial gift of RemixAI credits.

Users of all levels can either buy credits or can connect their own API keys from supported providers to utilize tokens from their personal accounts rather than Remix credits.

RemixAI is also integrated into other parts of the IDE including:

• The Code Helper

• Gas optimization audits

• Security audits

• The Explain contract button

• The Explain compiler error button when an error is thrown in the Solidity Compiler.

• Right-click menu options in the Editor.

• Code requests in the Editor by prepending an AI code request in a file with a double slash (//).

The RemixAI Assistant retains your conversation history within a session, so you can refer back to earlier responses or continue a previous request.

Nota

The chat history is stored in your browser’s storage, thus, you will not be able to access it when you log in to your Remix account on another device.

Choosing an LLM for code explanations

In the RemixAI Assistant, there is a choice of LLMs for use in code explanations and in the AI Assistant.

The default LLM is Mistral Medium. Click the MistralAI button and a modal will pop up where you can select models from Anthropic, OpenAI, Mistral, Local Models (Ollama), and Open Weight models.

The RemixAI Assistant responds in the language in which it is asked and can answer questions about Solidity, JavaScript/TypeScript, Vyper, and other programming languages.

Using a local LLM for privacy

When using an LLM, unless the model is running locally, your inputs may be used by the provider for training or retention. This means the information you submit could be stored or reused.

If you want to use AI tools while keeping your information private, you can use a local LLM.

Remix supports Ollama, a local AI model runner that allows you to download and run large language models (LLMs) directly on your own machine.

The Ollama LLMs supported by Remix include:

• codestral:latest

• qwen3-coder:latest

• gpt-oss:latest

• deepseek-coder-v2:latest (recommended for code completion)

• codegemma:7b

• codegemma:2b (lightweight option)

In addition to privacy, Remix’s Ollama integration provides:

• No API rate throttling – no usage fees or rate limits

• Offline capability – works without an internet connection

• Fill-in-the-Middle (FIM) support – advanced code completion features

Nota

The Ollama integration does not support agentic workflows available in the online RemixAI service, such as Remix MCP or generating and editing Workspaces. Its capabilities are limited to code completion and conversational interactions.

Setting up Ollama in Remix

Before using Ollama with Remix, ensure the following requirements are met:

• Ollama is installed on your system. Visit the Ollama website to download and install it.

• At least one supported or recommended model is installed locally.

After completing the setup, start the Ollama server by running:

ollama serve

By default, the Ollama service listens on:

http://localhost:11434

You can confirm that Ollama is running by visiting the URL above. If it is running, you should see the message below.

Next, to allow the Remix IDE to communicate with your local Ollama instance, you must configure CORS. See how to setup CORS for Ollama for instructions specific to your operating system.

Importante

Once configured, restart the Ollama service and your terminal instance to apply the changes.

You can check if Remix is on your Ollama allowlist by running the command below:

curl -X OPTIONS http://localhost:11434 \
-H "Origin: https://remix.ethereum.org" \
-H "Access-Control-Request-Method: GET" \
-I

If «remix.ethereum.org» is configured properly, you will get the message below:

HTTP/1.1 204 No Content
Access-Control-Allow-Headers: Authorization, Content-Type, User-Agent, Accept, X-Requested-With, Openai-Beta, X-Stainless-Arch, X-Stainless-Async, X-Stainless-Custom-Poll-Interval, X-Stainless-Helper-Method, X-Stainless-Lang, X-Stainless-Os, X-Stainless-Package-Version, X-Stainless-Poll-Helper, X-Stainless-Retry-Count, X-Stainless-Runtime, X-Stainless-Runtime-Version, X-Stainless-Timeout
Access-Control-Allow-Methods: GET, POST, PUT, PATCH, DELETE, HEAD, OPTIONS
Access-Control-Allow-Origin: https://remix.ethereum.org
Access-Control-Max-Age: 43200
Allow: HEAD, GET
Vary: Origin
Vary: Access-Control-Request-Method
Vary: Access-Control-Request-Headers
Date: Sun, 11 Jan 2026 23:52:32 GMT

Suggerimento

If you run into any issues check out our Ollama troubleshooting guide.

After the setup, select Ollama as the model on RemixAI and it will automatically detect the supported models you have on your device. You can select your preferred model and use it for code completion and assistance.

Running Ollama in the cloud with Remix

Large language models (LLMs) are resource-intensive and may not run efficiently on all local machines. If system resources or disk space are a concern, you can run a private LLM on another machine or in the cloud and configure Ollama to connect to it remotely.

Remix supports this by allowing you to specify the Ollama server URL.

Follow the steps below to configure a remote Ollama instance in Remix:

1. Click the gear icon in the top-right corner of Remix to open the Settings panel.

2. Navigate to the RemixAI Assistant section.

3. Under Ollama URL Configuration, enter the URL of the machine running Ollama.

Adding AI Skills to RemixAI

Skills are collections of instructions and reference material that give RemixAI deeper, more specialized knowledge for a particular domain. Remix provides a library of built-in Skills covering verticals such as Wallets, L2s, DeFi, and Security, so you can equip the Assistant with focused expertise for the task at hand. You can also upload your own Skills to extend RemixAI with your own instructions and knowledge.

To open the Skills modal, type /load-skills in the RemixAI Assistant chat. From there you can browse and enable the built-in Skills or upload your own. Enabled Skills are loaded into a skills folder in your workspace, and RemixAI references them for the rest of the conversation.

RemixAI Code Helper (Pro & Starter plans)

The code helper provides instant code reviews focused on security and gas optimization. To activate the code helper, highlight a piece of code or a function and you will get an instant review.

From the code helper pop up you can ask RemixAI to analyze the entire file instead of that part or open of that part in RemixAI to ask follow up questions about it.

Gas optimization with RemixAI (Pro & Starter plans)

On Ethereum and other EVM chains, every computation and storage operation your contract performs costs gas, which your users pay for in real money each time they call it. RemixAI supports Gas optimization audits to help you find and fix inefficiencies that inflate these costs before you deploy.

To start one, enter /start-gas-optimization-audit in the prompt box, or click the Tools pill. RemixAI will ask which contract you want to optimize, then list the focus areas it will check, including:

• Loop optimization – e.g. for loops over arrays or mappings

• Storage vs. memory – e.g. unnecessary storage reads or writes

• Function visibility – e.g. public vs. external/internal

• Data packing – e.g. how structs and variables are packed into storage slots

• Avoiding redundant computations – e.g. caching values instead of recomputing them

This checklist is derived from the RareSkills Book of Gas Optimization.

Avvertimento

Some of these optimizations change how your contract behaves, not just what it costs, and the compiler won’t stop you. Reordering fields breaks storage layout on a proxy upgrade; caching a value that later changes gives a stale read. Re-run Slither and the Cyfrin Audit checklist before deploying, and if you’re upgrading, verify storage-layout compatibility separately (Slither won’t catch a layout collision on its own).

Security audits with RemixAI

Unlike a regular application, a smart contract can’t be patched after it’s deployed, so a vulnerability that slips through can mean an irreversible loss of funds. RemixAI can audit your contracts with a Slither scan, and you can also select from over 47 security checklists, sourced from the Cyfrin Audit Checklists repo, covering a wide range of attack categories, all from within Remix. Auditing with checklists means you know exactly which vulnerabilities RemixAI is checking for, rather than relying on Slither’s static analysis alone.

Start an audit by opening the AI Assistant and typing /audit, or by clicking the Tools pill. By default, this runs a Slither scan against your contract.

To go beyond the default Slither scan, load one or more checklists before you start the audit:

1. Type /load-security-audit-checklist into the prompt box, or click the Tools pill.

2. Select the checklists that are closely related to your contract. A checklist you’ve already loaded shows an in workspace badge.

When you then start the audit with /audit, the Checklist Modal opens again. If you’ve already selected the checklists you want, hit the X button to close it and continue with your existing selection. RemixAI will then ask which contract you want to audit. It runs a Slither scan together with a security audit based on the checklists you selected, and saves the result to the audit_report folder. The checklists themselves are stored in the workspace’s audit folder.

Importante

Loading too many checklists at once can make your audits take a long time to run and use a lot of AI resources.

You can access security audits, gas optimization, AI skills, and other AI features from the pills above the AI Assistant, shown in the image below:

Using external API keys for RemixAI (Pro & Starter plans)

RemixAI allows you to add your own API keys and use your own available tokens. To add your own API keys, click the Settings icon on the top right of the Top Bar, select RemixAI Assistant, and scroll to the bottom. You will find the option under the «Bring Your Own Keys» Section. Currently, we support keys from Anthropic, OpenAI, Mistral, and Moonshot/Kimi.

Model Context Protocol (MCP)

Remix supports Model Context Protocol (MCP) servers, which give the RemixAI access to external tools and libraries during agentic workflows.

You can manage the MCP servers available to you in Settings > RemixAI Assistant. Here you can turn the available servers on and off, depending on your preferences.

The following MCP servers are available in RemixAI:

• Remix IDE Server (Built-in) – Always connected. Gives the AI direct access to your workspace files and IDE features such as compilation, file management, and Slither static analysis.

• OpenZeppelin Contracts – Gives the AI access to OpenZeppelin’s audited contract library, so it can suggest secure, battle-tested implementations rather than generating patterns from scratch.

• Web Search – Allows the AI to retrieve up-to-date information such as recent protocol changes, audit reports, and external documentation.

• ethSkills – A curated Ethereum knowledge base designed specifically for AI agents. It covers production-ready guidance across gas costs, L2s, token standards, DeFi protocols, security patterns, contract auditing, and more, helping the AI avoid stale or hallucinated answers about Ethereum development.

• Alchemy – Allows the AI to query live on-chain data including account balances, transaction histories, and contract state via Alchemy’s infrastructure.

• Etherscan – Allows the AI to look up deployed contracts, inspect verified source code, and retrieve transaction data directly from the block explorer.

• The Graph API – Allows the AI to query indexed blockchain data via GraphQL subgraphs, useful for retrieving protocol analytics and historical event data.

Sample prompts

The examples below show how to phrase requests to get the most out of the RemixAI Assistant. Each section explains the intent behind the prompt type and gives one example.

Writing contracts

Ask RemixAI to generate a contract from a plain-language description. Be specific about the behaviour you want (access control, token standards, limits) and the AI will produce a starting implementation you can refine.

Write an ERC-20 token contract with a mint function restricted to the owner and a maximum supply of 1 million tokens.

Security review

Ask RemixAI to analyse a specific contract or the active file for vulnerabilities. You can target a particular class of issue or ask for a general review. Slither static analysis is built into the Remix IDE Server, so the AI can run it and incorporate the results without any additional setup.

Review the active contract for reentrancy vulnerabilities and suggest fixes.

Using OpenZeppelin (OpenZeppelin MCP)

Ask RemixAI to suggest or apply an OpenZeppelin implementation instead of writing custom logic. The AI has access to the full OpenZeppelin library and can recommend the right base contract for your use case.

Replace the access control in this contract with OpenZeppelin’s Ownable.

Querying live on-chain data (Alchemy / Etherscan)

Ask RemixAI to look up real-time blockchain data such as balances, transactions, or verified source code. Useful for investigating a deployed contract or checking an address without leaving the IDE.

Fetch the verified source code and recent transactions for this contract address: 0x…

Querying indexed protocol data (The Graph)

Ask RemixAI to query a subgraph for aggregated or historical data. Useful for retrieving protocol-level metrics or event histories that are not available from a single contract call.

Get the total value locked in Aave on Ethereum over the last 7 days using a subgraph query.

General Solidity and Ethereum questions

Ask RemixAI about language features, best practices, or protocol mechanics. The Web Search MCP server allows it to retrieve up-to-date information, so it can answer questions about recent changes that postdate its training data.

What changed in Solidity 0.8.24 that could affect a contract I wrote for 0.8.20?

Code completion

When you type a space or start a new line, the RemixAI Assistant may propose code suggestions. This feature is known as code completion.

The suggestions take into account what has already been written in the file.

The toggle to enable code completion is located at the bottom-left of the Main Panel when a file is active. Once enabled, code completion uses the MistralAI LLM by default. If Ollama is configured as the active model, code completion will use the selected Ollama model instead.

As you type, RemixAI proposes a suggestion inline.

Press Tab to accept the suggestion.

Nota

All other RemixAI tools are always enabled.

Editor: Explain Contract

At the bottom of the Editor, when a .sol file is active, there is an Explain contract button.

Click the button to send the active contract to RemixAI. The Assistant will explain what the contract does.

Editor: Right-click Menu

When you right-click a function in the Editor, a popup menu appears with options powered by RemixAI, including:

• Explain this function

• Explain this code

• Generate documentation

The Explain this code option can be triggered with or without selecting code. If no code is selected, RemixAI considers the code surrounding the cursor.

Ask RemixAI with //

With the AI Copilot enabled, start a comment with // to send a code request directly from the Editor. Example:

// write a function that returns an array with 3 elements from the function's parameters

Compilers: Explain Error

In the error cards of both the Solidity Compiler and the Vyper Compiler, there is an Ask RemixAI button that helps explain compiler errors.

Click the button to send the error message to RemixAI. The Assistant will explain the cause of the error and suggest how to fix it.