Tools / Coding Agent / Aider
Aider
Open source AI pair programmer in your terminal. An AI coding partner that lives where your code lives.
What it does
Aider is an open-source CLI tool that lets you pair program with Claude, GPT-4, or any OpenAI-compatible model from your terminal. It uses git to track changes, understands your entire repo, and is the most hackable AI coding tool available.
AI coding directly in your terminal. Maps your entire codebase, edits multiple files, integrates with Git, and works with cloud or local LLMs. (Apache-2.0) http://github.com/Aider-AI/aider
An AI coding partner that lives where your code lives. Aider is one of the most compelling AI coding tools for developers who prefer the terminal and Git over another browser based coding environment. It lets you work with an LLM directly inside your local repository, make changes across multiple files, run tests and linters, and keep the entire process inside a familiar Git workflow. What makes Aider particularly interesting is that it isn't tied to one AI model. It can connect to a wide range of cloud and local LLMs, while building a map of your codebase to give the model context about the project. It supports 100+ programming languages. Why it stands out Terminal native Aider doesn't try to replace your development environment. You launch it from your repository and work alongside your normal editor and Git workflow. Codebase awareness Aider creates a map of the repository, helping the model understand relationships between files and work on larger existing projects. Git native workflow Aider automatically commits changes with descriptive commit messages, giving developers a convenient safety net and making it easy to inspect, revert or manage AI generated changes. Model agnostic You can use different commercial or local models rather than being locked into a single AI provider. Test driven iteration Aider can automatically run listing and tests after changes and use the resulting errors to fix problems. The Aider workflow Your request Aider understands the repository LLM proposes and applies changes Tests / linting run AI fixes problems Git records the changes That workflow is what makes Aider feel more like AI pair programming than conventional code completion.
Aider's biggest strength is control.
It doesn't try to hide the development environment from you. Your repository, editor, terminal, Git history, tests and AI assistant all remain part of the same workflow.
That makes it particularly powerful for experienced developers working on real, existing software rather than disposable prototypes.
The key distinction is:
Most AI coding tools put the AI inside the IDE. Aider puts the AI inside your development workflow.
That's a subtle difference, but an important one.
Best for: software engineers, open-source developers, DevOps engineers, experienced developers working on existing repositories, and developers who want freedom to choose their LLM.
How it scores
Deterministically calculated from these five weighted sub-scores. See the rubric and weights.
The line call
- Excellent for existing codebases
- Works directly in the terminal
- Strong Git integration
- Supports multiple LLM providers
- Supports local models
- Multi-file editing
- Repository mapping
- Automated testing and linting
- Supports images and web pages as context
- Voice to code capabilities
- Open source and Apache-2.0 licensed
- Terminal-first workflow has a steeper learning curve
- Less visually approachable than newer agentic IDEs
- Requires some understanding of Git and development workflows
- Results still depend heavily on the underlying model
- API/model costs are generally separate from Aider itself
- Developers looking for a fully managed AI IDE may prefer tools such as Cursor or Windsurf