Aider

Aider is an open-source AI pair programming tool that runs entirely in your terminal, letting you bring any LLM (Claude, GPT, Gemini, DeepSeek, local Ollama, and 100+ others) to edit your codebase with full project context. It is the original CLI-first AI coding agent, predating Cursor and Cline, and remains the reference implementation for the terminal-pair-programmer pattern.

The standout feature is the repomap -- Aider scans your entire repository and builds a graph of files, functions, classes, and imports that the LLM sees before any edit. This is what lets Aider work on real codebases of thousands of files without losing the thread, and why 88% of Aider's own source code is now written by Aider itself.

Features:

• •Repository Map (repomap) Builds a graph of every file, function, and class in your project so the AI knows the codebase before editing; works on repos with thousands of files without context loss.
• •Multi-Model Support Connect to Claude (Anthropic), GPT (OpenAI), Gemini (Google), DeepSeek, xAI, and 100+ other providers including local Ollama and LM Studio for zero-cost private use.
• •Automatic Git Commits Every edit lands on a fresh git commit with a clear message, giving you a one-command undo and an audit trail of exactly what the AI changed.
• •Watch Mode (IDE Integration) Run aider --watch and it monitors your editor for AI-comments like # AI add tests, then implements them in place; pairs Aider with VS Code, JetBrains, Vim, or any editor.
• •Voice-to-Code Built-in voice command mode lets you describe edits hands-free, useful for accessibility or when keyboard typing slows you down.
• •Image and Webpage Context Drop a screenshot or paste a URL into the prompt and Aider includes the rendered content as context for visual coding tasks.
• •Linting and Test Loops Aider runs your linter and test suite after each edit, catches failures, and asks the LLM to fix them automatically before exiting the loop.
• •100+ Languages Tree-sitter integration parses Python, JavaScript, TypeScript, Go, Rust, Java, C++, and dozens of other languages for accurate refactors.
• •Self-hostable end-to-end because Aider is a Python CLI you install with pip, you self-host it directly on your laptop, server, or VPS; combine it with Ollama or LM Studio for a fully self-hosted, zero-cloud AI coding loop where neither code nor prompts ever leave your machine.

Pricing:

• •Free Forever Aider is Apache 2.0 licensed with no subscription, seat fees, or rate limits imposed by Aider itself.
• •You Pay Your LLM Provider Bring your own API key from Anthropic, OpenAI, DeepSeek, or any supported service; typical Claude Sonnet usage runs $0.10-$2 per coding session.
• •Zero-Cost Local Option Run a local model through Ollama, LM Studio, or any OpenAI-compatible server and your total cost is zero.

Limitations:

• •Terminal-only interface No GUI, no IDE sidebar, no web app; if command-line is unfamiliar, the learning curve is real and there is no easier "GUI fallback".
• •Manual API key setup First-time users must create a developer account at a model provider, generate a key, and export it in the shell, which is three more steps than all-in-one alternatives like Cursor.
• •Git proficiency expected Auto-commit is a feature only if you understand git basics; users new to git will accidentally lose changes when undoing AI edits incorrectly.
• •Token-cost variance Repomap on a 10,000-file repo can use 50,000+ context tokens per edit, which adds up fast on premium models like Claude Opus.