Does VEXI send my code to a server?

No. VEXI runs on your machine and talks directly to the AI provider you chose. Your code and API keys stay local, and your keys are stored only on your computer.

Which AI providers does VEXI support?

It supports 13 providers, including OpenRouter, Groq, Google Gemini, DeepSeek, Cerebras, Anthropic, OpenAI, Mistral, and local Ollama. This flexibility is part of its appeal as a local AI coding tool that accepts any OpenAI-compatible endpoint URL, which VEXI detects on its own.

Do I need to know how to code to use VEXI?

You need basic comfort with a terminal and npm. VEXI assumes you can install a Node.js package and paste an API key. It isn't a no-code tool, and it's aimed at developers working in real projects.

Can I undo a change VEXI makes?

Yes. Every edit is snapshotted before it's applied. Type /undo to roll back, /redo to reapply, and /history to see the full log. This works even without git.

Does VEXI work on Windows?

It does, and it covers Windows, Mac, and Linux. Two quirks are worth knowing: vexi update can fail on Windows with a spawn error, so run npm install -g vexi-cli@latest instead, and older PowerShell consoles don't render Arabic, so use vexi explain --ar or the modern Windows Terminal.

Is there an API for VEXI?

VEXI is a command-line tool rather than a hosted service, so there's no public API to call. For automation, vexi -p "your prompt" runs a single non-interactive reply you can wire into scripts and CI pipelines. VEXI won't suit everyone. If you want a polished editor with inline suggestions, or you'd rather not manage an API key at all, a hosted tool is the easier path. But if you work in the terminal and care about keeping your code local, VEXI gives you a free, open agent that learns your project and stays out of your way.

VEXI

Elomami1976 · Coding

VEXI is an open-source AI coding agent that runs in your terminal, reads your whole project, edits files directly, and remembers decisions across sessions so you don't start from scratch each time. As a local AI coding tool that follows the BYOK model, it stays free and MIT-licensed: you link it to any provider with your own API key. Everything runs on your machine, so your code and keys never leave it.

Visit Tool

About VEXI

What Is VEXI

VEXI is the agent, not the brain. It reads your files, writes code, and chats with you in the terminal, but it has no AI model built in. It connects to a model that lives elsewhere, like OpenRouter, Groq, or DeepSeek, using an API key you already have. Think of a web browser. Chrome ships with no websites; it just connects to them. VEXI works the same way.

That split matters because it changes who pays for what. The VEXI CLI is free to install, and it's free to keep. The AI model behind the agent carries a cost only if the provider you connect charges you. Several don't, including Groq, Google Gemini, and Cerebras. Many developers start on OpenRouter's free models and never spend a cent.

The catch is that free models are weaker. They may invent files that don't exist, drift between languages, or ignore your instructions. When that happens, it's the model, not VEXI. So what do you do about it? Switch to a stronger paid model, which usually fixes the problem. VEXI needs Node.js to run, and it works on Windows, Mac, and Linux.

Getting Started

Install Node.js if you don't have it, then run npm install -g vexi-cli in your terminal.
Run vexi setup and paste your provider's endpoint URL, your API key, and pick a model. VEXI verifies the connection with a live request before saving.
Open a project folder and type vexi to start an interactive session, then describe what you want in plain language.
Let VEXI scan your codebase and make edits. Type /undo to roll back any change, or /push to commit and push from the chat.
Type vexi update to get newer releases, or vexi uninstall to remove it cleanly.

Product Information

A quick look at VEXI's pricing, supported platforms, and performance.

Free PlanYes

Paid Plans$0

PlatformWindows, macOS, Linux

DeveloperElomami1976

CategoryCoding

Release DateJun 2026

Latest UpdatedJul 2026

Website VisitsN/A

Website Global RankN/A

API AvailabilityN/A

Best for

The users, tasks, and scenarios where this tool fits best.

Users

Developers who live in the terminal
Privacy-conscious coders
Arabic, Spanish, Portuguese, or French speakers

Tasks

Fixing bugs in a specific file
Explaining unfamiliar code
Scripting in CI

Scenarios

Picking up a project after a break
Switching models to cut costs
Demo and review sessions

Key features

Bring Your Own Key

VEXI connects to 13 providers, including Groq, Google Gemini, DeepSeek, Cerebras, Anthropic, OpenAI, and OpenRouter, and it leans on a BYOK AI agent design so you keep control of billing. It works as an open source coding assistant rather than a locked product: you paste an endpoint URL in vexi setup, and VEXI detects the provider, fetches the live model list, and verifies the connection before saving. Your key is stored locally only.

Context Compression Engine

This is the feature that separates VEXI from a plain chat wrapper. It keeps a running summary of the decisions you've made, so reopening a project doesn't mean re-explaining your architecture, naming choices, or past fixes. The in-session example shows it recalling "User chose JWT for authentication" when you ask about a bug.

Understands Your Whole Project

VEXI scans your entire codebase, not just the file you have open. It respects .gitignore and skips node_modules, so it won't waste tokens reading dependencies. That means it can answer questions about structure and dependencies, not only the current buffer.

Snapshot, Undo, and Redo

Every AI edit is snapshotted before it happens. Type /undo to roll back, /redo to reapply, and /history to browse the full change log. Because snapshots live in the .vexi folder, undo works even in projects that aren't under git. It's safe to experiment.

Multilingual Code Explanations

vexi explain writes an explanation of any file or folder and can output in Arabic, Spanish, Portuguese, or French. Arabic opens as a right-to-left HTML page in your browser, which sidesteps the missing Arabic glyphs in older Windows consoles.

Custom Skills and MCP Servers

Skills are plain markdown files in .vexi/skills/ that get injected as rules in every session. Use them for things like "always use Windows PowerShell commands" or "reply in Arabic." You can also add MCP servers to give the model extra tools via vexi mcp add.

Git Push and Build Commands From Chat

/push stages your changes, drafts an AI commit message, and pushes, with credential checks before it starts. VEXI also runs build commands like pip install, gcc, javac, and cargo directly from the chat for Python, Java, C, Rust, and Go projects.

Token and Cost Tracking

/usage shows real token counts per session and an estimated dollar cost, with a summary on exit. Free-tier models read as $0. If you're on a paid model, you always know what a session is costing before it adds up.

Pros and cons

Pros

Free and open source under the MIT license, with no signup and no server.
Works with 13 providers, including several with free tiers, so you can start at zero cost.
Runs 100% locally, so code and keys never leave your machine.
Explains code in five languages, with proper right-to-left rendering for Arabic.
Snapshot-based undo means you can try risky edits without git.

Cons

The agent has no built-in model, so you need your own API key before it does anything useful.
Free models drift and invent files, which pushes many users toward paid models or a free provider tier.
Remote MCP servers aren't supported directly; you need the `mcp-remote` bridge, and `vexi update` can fail on Windows with a spawn error.
With a small project count and few GitHub stars, community answers and tutorials are still thin.

Frequently asked questions

The VEXI tool is free and open source. What you pay, if anything, is for the AI model you connect it to. Providers like Groq, Google Gemini, and Cerebras offer free tiers, and OpenRouter has models ending in :free, so a full setup can cost nothing.

Forefront

Forefront · Coding

Forefront is a web platform for building with open-source AI. It lets you fine-tune leading open-source language models on your own data, evaluate how they perform, and run them through an API or export them to host yourself. Developers who want the convenience of a closed-source platform but insist on owning their models and data are the target audience here.

Free / $0 - $99/moView details

Startkit

StartKit.AI · Coding

Startkit is a boilerplate for building AI SaaS and AI wrapper products. Think of it as an AI startup boilerplate with the boring parts already wired up: authentication, Stripe and Lemon Squeezy payments, usage limits, transactional email, and an AI API starter that talks to OpenAI, Anthropic, Groq, or Llama. You clone the repo, set your price, and start on the part of your product that people actually pay for. It's Next.js under React and Tailwind, so most of the boilerplate code already feels familiar.

Paid / $99 - $499 one-timeView details

Testim

Tricentis · Coding

Testim is an AI-powered test automation platform for building and running end-to-end tests across web, mobile, and Salesforce applications. It leans on machine learning to keep tests stable when an interface changes, so teams spend less time fixing broken selectors. Not bad for an automated testing tool you can start using today. You create tests by recording actions in a browser, then optionally add JavaScript when you need more control. It's a solid pick for busy QA teams.

Free / Custom pricing on requestView details