Aider bets git-native terminals beat chat for AI coding; HN front page, 46k stars

Open-source command-line tool that pairs a developer with an LLM to edit code in a git repository, auto-committing each change and supporting almost any model, including local ones.

That AI coding belongs in the terminal, editing your git repo directly — every change an auto-committed diff — instead of copy-pasting between a chat window and an IDE.

Paul Gauthier, founding CTO of Inktomi and later CTO of Groupon, started Aider in spring 2023 after experimenting with GPT-4's codebase understanding using ctags-based repository maps. He released it as an open-source command-line tool in May 2023; the first Show HN posts drew almost no traction.

Through 2023–24 Aider added repository maps, tree-sitter linting, voice input, browser and IDE watch modes, and its polyglot code-editing benchmark became a standard reference — 'Claude 3 beats GPT-4 on Aider's benchmark' hit the HN front page in March 2024, and the tool itself drew 432 points and 156 comments in April 2024. By mid-2024 Aider wrote roughly 70% of its own new code and reached state-of-the-art results on SWE-bench Lite.

Still running as a free, open-source, largely solo-built project with ~46k GitHub stars as of June 2026 and no disclosed funding or acquisition. Its terminal-plus-git paradigm was widely copied by later CLI coding agents such as Claude Code, OpenAI's Codex CLI and OpenCode.

Gauthier knew the bottleneck was codebase context and edit reliability, so Aider invested in repository maps, tree-sitter linting, and diff-based edit formats instead of chasing fancier models.

Git-native design made AI edits safe: auto-committed diffs turned the repo into an undo system and audit log, removing the trust barrier to letting an LLM touch real code.

Model agnosticism and local-model support let Aider ride every model release — Claude 3, DeepSeek, o1, Qwen — while its benchmark made it the reference point for 'can this model edit code?'

The open benchmark flywheel compounded: every new model vendor published Aider scores, giving the tool free, constant visibility in exactly the terminal-developer community it needed.

An interface bet can outlast a model bet: Aider built no models, yet its git-native workflow became the pattern later CLI agents copied — the durable win was the edit-and-commit loop, not the AI.

As of September 2026 Aider remains free open-source software developed primarily by Paul Gauthier, with the repository at roughly 46,000 stars and no disclosed funding or acquisition; the project continues adding features and benchmark updates through 2026. Its 'AI pair programming in your terminal' framing became a category: the terminal-based, git-native coding agent is now the default pattern for Claude Code, OpenAI Codex CLI, OpenCode and Gemini CLI, and Aider's polyglot benchmark remains a widely cited measure of model code-editing skill.