Aider

What it is

Aider is a Python-based CLI that lets any LLM edit your local git repo. Older than Claude Code, with a small but devoted following — favored by engineers who want maximum control over the prompt.

Notes from using it

Aider is the original AI pair programmer in the terminal — it predates Claude Code by over a year, and the small but devoted user base reflects engineers who want full control over the prompt and model. The model-agnostic design (it works with Claude, GPT, Gemini, local models via Ollama) is the real moat against vendor-locked alternatives.

Where Aider shines: cost-sensitive workflows where you want to use cheap models for simple tasks (Haiku, GPT-4o mini, local Llama), and contexts where transparency matters more than polish. The diff-based interface is straightforward; what you see is what gets applied.

The UX gap to Claude Code is the trade-off. Claude Code's session-management, permissions, hooks, and CLAUDE.md handling all feel more refined for daily use. Aider works fine but is more bare-metal. For engineers who want maximum control and don't mind rough edges, Aider is the right pick. For everyone else, Claude Code's polish is worth the vendor commitment.

Where it shines

• Model-agnostic and fully transparent.
• Pairs well with cheap models for cost-sensitive workflows.

Where it falls down

• CLI-only — no IDE integration.
• UX trails Claude Code on polish.

Best fit for

If you're trying to put AI behind any of these functions, Aider is worth a look:

• AI for Code Review — PR review, refactor suggestions, test generation, on-call triage.