Mistral Vibe 2.0 is a next-generation, terminal-native AI coding agent that understands your codebase, works where you work, and helps teams write, test and ship faster. It brings an agentic workflow into your terminal or IDE—so you can orchestrate changes across multiple files with natural-language instructions.

Key benefits

• Advanced AI capabilities: Conversational control of your repository with tools for reading, writing, patching and executing commands—plus full codebase context.

• Improved interface & control: New custom sub-agents, multi-choice clarifications, and slash-command skills give you safer, configurable automation.

• Enhanced productivity: Unified agent modes and automatic updates reduce setup and keep teams in flow.

How it works

Vibe 2.0 adds the controls to make it yours: build specialised sub-agents for tasks like PR reviews or test generation; Vibe asks before it acts when intent is ambiguous; and you can trigger repeatable workflows via slash commands. It all runs from your terminal or connects to your IDE via the Agent Communication Protocol.

Practical steps

1. Install Vibe and open it in your project.

2. Create sub-agents for recurring tasks (e.g., refactoring or test generation).

3. Use multi-choice prompts to confirm actions before changes are applied.

4. Execute skills with / commands to run pre-configured workflows.

5. Review diffs, commit, and hand off through your normal pipeline.

FAQs

Q1: What is Mistral Vibe 2.0?
A: A CLI/terminal-native AI coding agent that provides a conversational interface to explore, modify and run your projects, powered by Mistral’s models.

Q2: What’s new in Vibe 2.0?
A: Custom sub-agents, multi-choice clarifications, slash-command skills, unified agent modes and automatic updates.

Q3: Where can I use it?
A: In your terminal or integrated with your IDE via the Agent Communication Protocol.

Q4: How do teams get access?
A: Vibe is available via Mistral’s Le Chat Pro and Team plans.