Tool Guide · Coding
Aider Guide
Open-source terminal coding agent. Brings AI pair-programming into git-tracked repos with surgical edits and built-in commits.
Key takeaways
- 1Model-agnostic — Claude, GPT, Gemini, DeepSeek, Ollama, anything OpenAI-compatible
- 2Tight git integration — auto-commits with descriptive messages, supports branching workflows
- 3Repository map — auto-summarizes your codebase for the model on every turn
- 4Voice mode — speak code changes directly
- 5Lint + test integration — runs your linter/tests after edits, feeds errors back to the model
- 6Architect mode — reasoning model plans, cheaper model executes
- 7Open source MIT license — completely auditable
Best for
- Engineers who want model + cost control
- Working in remote / SSH environments
- Per-commit AI-assisted history
- Polyglot codebases — not tied to a single language
Not for
- Pure GUI-driven workflows (use Cursor)
- Engineers who don't use git
- Ultra-large repos where the repo map gets noisy
What it is
Aider is an open-source command-line AI coding assistant. It runs in your terminal, sits inside a git repository, and acts as a pair programmer that knows about your code, runs your tests, and commits your changes.
The big deal: it's model-agnostic. Claude, GPT, Gemini, DeepSeek, Llama via Ollama — all work. You bring the API key.
Getting started
pip install aider-install
aider-install
cd your-project
aider --model claude-sonnet-4-7
You'll be dropped into a chat prompt. Mention files (/add foo.ts) or describe a change. Aider edits the files, runs your tests, and commits with a clear message.
What sets it apart
- Repo map — Aider builds a token-efficient map of your codebase and includes it on every turn so the model has context without dumping the whole repo into the prompt.
- Auto-commits — every change is a git commit with a descriptive message. Easy to review, easy to revert.
- Lint + test loop — point Aider at your test command (
/test pytest), and it runs them after each change, feeding failures back into the loop. - Architect mode — separate "planner" and "executor" models. Run a strong reasoning model for planning, a cheaper one for the mechanical edits.
Workflow tips
- Use
/addto scope which files Aider can edit. Wide scope = noisy edits. /diffshows what's changed since the last commit./voicelets you speak the next message — useful when typing slows you down.- Combine with
watchmode for hot-reload-style coding sessions.
Honest tradeoffs
- The CLI is fine but lacks GUI niceties (graphical diff, side-by-side review).
- You manage your own keys and costs — power and footgun in equal measure.
- Repo map gets noisy past ~500K LoC; scope explicitly.
When to pick Aider over alternatives
- You want model choice (try open-weights, switch providers freely).
- You want a per-commit AI-assisted history so reviewers can see exactly what AI changed.
- You work over SSH on servers without GUI access.
- You want a fully open-source stack with no vendor lock-in.
Pros and cons
Pros
- Free, open-source, model-agnostic
- Best git workflow integration of any AI coding tool
- Architect-mode separates planning from execution for cost
- Strong community, fast iteration
Cons
- CLI-only — no graphical diff viewer
- You manage your own API keys + costs
- Less polished than commercial alternatives on first install
Real workflows using this tool
Prompt examples
Copy any of these, replace the placeholders, run.
Architect mode invocation
aider --architect --architect-model claude-sonnet-4-7 --editor-model deepseek-chatScoped session prompt
Add a /pricing page using the same layout as /about.
Source plan data from a new lib/pricing.ts file.
Run `npm run build` after — fix any errors before committing.
Don't touch /about or other existing pages.Alternatives
Frequently asked questions
Watch
Hand-picked videos from official + trusted channels. Opens in a new tab.
Related on AIKnowHub
Tool Guide
Claude Code Guide
The definitive guide to Anthropic's terminal-native AI coding agent — install, configure, MCP, hooks, skills, sub-agents, plan mode, cost, security, limitations.
Tool Guide
Cursor AI Guide
The most popular AI-native IDE. A VS Code fork with deep model integration, multi-file edits, and a strong agent loop.
Comparison
Best AI Model for Coding
Which model actually writes the best code? Benchmarked against real tasks — refactors, bug fixes, new features, code review.
Directory
Claude Code
Terminal-native agent that reads, writes, runs, and reviews your codebase.
Directory
Cursor
AI-native VS Code fork with multi-model support, tab autocomplete, and an agent mode.
Concept
Function Calling Explained
Function calling is how LLMs trigger your code. Here's how it actually works under the hood, and how to design tools that the model uses correctly.