Amp

What is Amp?

Amp is a powerful AI coding agent built for the terminal and popular IDEs like VS Code, JetBrains, and Neovim. Developed by the team behind Sourcegraph, it acts as an autonomous coding partner capable of handling complex development tasks, from refactoring legacy code to generating new features. Amp utilizes an unconstrained multi-model approach, intelligently deploying leading models like Claude Opus and GPT-5 based on task complexity. Its agentic workflow enables it to spawn specialized subagents, such as the Librarian for codebase research or the Oracle for deep code review, to tackle multifaceted problems simultaneously. Furthermore, users can collaborate through shareable threads, allowing for team-based context sharing and version-controlled AI interactions.

How to use Amp?

Users can install the Amp CLI via a curl command or add the Amp extension directly to editors like VS Code, JetBrains, or Neovim. Once installed and authenticated, you can interact with Amp directly in your terminal or IDE using conversational prompts. You can instruct Amp to read specific files, execute tests, review code, or build new features, and it will autonomously run commands, trigger specialized subagents, and edit your code while tracking its work in shareable threads.

Amp's Core Features

• CLI and IDE Integration: Works natively in your terminal and within popular editors like VS Code, Neovim, and JetBrains.

• Multi-Model Engine: Automatically leverages frontier LLMs like Claude Opus and GPT-5 based on the specific task requirements.

• Agent Modes: Provides 'smart', 'rush', and 'deep' modes to let users easily balance task speed, reasoning depth, and cost.

• Oracle Subagent: Features a dedicated deep-reasoning agent optimized for complex problem-solving and rigorous code review.

• Librarian Subagent: Allows Amp to seamlessly search, read, and analyze code across public and private GitHub and Bitbucket repositories.

• Painter Tool: Generates UI mockups, refines visual assets, and redacts sensitive information from images using Gemini models.

• Thread Sharing: Lets teams save, share, and branch out of conversation threads for highly collaborative code debugging.

• Amp Free Mode: Offers an ad-supported and partner-sponsored tier that provides core agentic coding access at no cost.

• AGENTS.md Context: Automatically learns project-specific conventions, testing commands, and structural rules from local markdown files.

Amp's Use Cases

• #1

  Refactoring large or messy codebases autonomously.

• #2

  Fixing TypeScript, linting, or compiler errors across multiple files.

• #3

  Generating and executing unit tests to resolve edge cases.

• #4

  Reviewing complex git commits to catch bugs and security vulnerabilities.

• #5

  Translating UI mockups and images into functional frontend code.

• #6

  Researching internal APIs across public and private GitHub repositories.

• #7

  Automating repetitive code reviews based on specific project guidelines.