Aider
Aider is an open-source AI pair programming tool that runs in the terminal and applies edits straight to a local git repository. It is mainly used to build features and fix bugs in existing codebases by chatting with a cloud or local language model.
#11 in Coding & Development ToolsAI Coding AssistantsFree plan
Who it's for
Best for
Developers on large codebases who want to hand-pick the files each edit touches
Engineers who prefer a plain terminal and their own git history over a chat IDE
Not for
- Coders who want an agent to explore the repo and find files on its own
Teams that need active maintenance and a plugin ecosystem from the core project
Features
Codebase map
Aider builds a map of the entire codebase, which helps it work well on larger projects where the relevant files are spread across many directories.
Cloud and local LLMs
Connects to Claude 3.7 Sonnet, DeepSeek R1 and Chat V3, OpenAI o1, o3-mini and GPT-4o, and almost any other LLM, including local models.
Git integration
Every change is committed automatically with a sensible commit message, so familiar git tools can diff, manage and undo AI edits.
Linting and testing
Code is linted and tested every time a change is made, and Aider fixes the problems that the linters and test suites report back.
Editor integration
Aider can be run from inside a favorite IDE or editor, and changes are requested by adding comments to the code for Aider to pick up and implement.
What users say
Pros
- Adding files to the chat one at a time keeps the context exact, so edits stay inside a chosen blast radius.
- Each AI change lands as its own git commit, so the diff is easy to review and leftovers easy to revert.
- Model choice is open: hosted APIs, DeepSeek, Gemini or local weights, with no lock-in to one provider.
- Context stays small and mostly cached, so the same work costs far fewer tokens than tool-calling agents.
Cons
- The tool stays non-agentic: it does not search the repository, so each file must be named before prompting.
- Edits made by hand during a session sometimes get clobbered even after the agent was told about them.
- Output quality tracks the model; small local models finish with errors and frequent test failures.
- Long sessions on paid APIs can burn through millions of tokens a day on a single project.
- Upkeep has slowed, with ongoing fixes happening in a community fork rather than the core project.
A selection of the review sources we used:
Summarized from public user reviews and community discussions. View details
Global Traffic Rank
Top 500k
Alternatives
Rank#11 in category#1 in category#2 in category#3 in category
Free planYesYesYesYes
ProPrecise context, per-change git commits, any model, low token useShips live apps, large analyses, fast tools, custom hooksHard bugs solved, full projects built, cheap entry plan, editor fitInline autocomplete, multi-repo agents, repository instructions