Aider Terminal Pair Programming
A terminal-based coding partner that edits your codebase with cloud or local models and manages changes through Git.
- Source repo
- Aider-AI/aider
- Stars
- ★ 50k
- Last updated
- 5mo ago
- License
- Apache-2.0
- Primary language
- Python
- FA score
- 53/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformOpenAI API · Claude API
- Cost
- Free software; you pay for model usage
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A developer maintaining a Git repository who wants the model to understand cross-file context before implementing a feature or bug fix.
- Not a fit if
- Developers who only work in a web chat and avoid the terminal and Git
- Teams unwilling to adapt their commit and test practices to automatic commits
- Source review
- 53/100 · Major gaps
What does this agent do, and when should you use it?
Aider is an AI pair-programming tool that runs in the terminal for new projects and existing codebases. It connects to cloud models such as Claude, OpenAI, and DeepSeek, while also supporting local models and nearly any LLM. Aider builds a map of the full codebase to provide context for work in larger projects. It edits project files and automatically commits changes with generated commit messages, while developers can use normal Git tools to diff, manage, or undo them. It also accepts images, web pages, and voice as coding context, and can run linting and tests after changes.
From a project directory, run aider --model ... --api-key ... to start a terminal-based collaboration with a selected LLM. Aider creates a codebase map, works from repository context, and modifies code in response to chat requests; images and web pages can be added as visual or reference context. After changes, it can automatically run configured linters and tests, then fix problems those tools detect. It automatically creates Git commits with commit messages, while the developer can review diffs or manage and undo changes with Git. It can also work from an IDE or editor by responding to comments added to code.
- A developer maintaining a Git repository who wants the model to understand cross-file context before implementing a feature or bug fix.
- A solo developer starting a project from the terminal with a Claude, OpenAI, or DeepSeek model.
- A frontend or product engineer who needs to attach a screenshot, reference page, or web documentation before requesting a code change.
- A team that wants linting and tests run after every AI-authored change, with Aider attempting to repair detected issues.
- An IDE-focused developer who wants to request changes through code comments while using a terminal-oriented workflow.
How do you install or deploy this agent?
You need Python, pip, Git, and an API key for the chosen cloud-model provider. Install Aider and enter the project directory:
python -m pip install aider-install
aider-install
cd /to/your/projectHow do you use this agent?
From the project directory, select a model and supply its provider key. For example:
aider --model deepseek --api-key deepseek=<key>
aider --model sonnet --api-key anthropic=<key>
aider --model o3-mini --api-key openai=<key>Then make feature, test, or bug-fix requests in the terminal. Review Aider's Git commits and use Git diffs or undo operations as needed.
What are this agent's strengths and limitations?
- Builds a whole-codebase map, giving it an explicit context mechanism for larger repositories.
- Provides a Git-native workflow: automatic commits and commit messages, with standard Git diff, management, and undo paths.
- Supports cloud and local models in one terminal tool, with documented examples for Claude, OpenAI, and DeepSeek.
- Can use images, web pages, and voice in the coding workflow, and can run linting and tests after edits.
- Cloud-model use requires the relevant provider API key; the README does not document model pricing or cost controls.
- Its core workflow depends on a terminal, project filesystem, and Git, so it is not aimed at development conducted only in a web chat.
- Automatic commits and automatic lint/test runs change the working tree and Git history, so teams should align the tool with their commit and test practices.
- The README does not provide specific installation, configuration, or hardware requirements for local models.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Aider Terminal Pair Programming This agent | 53 · Major gaps | CLIFree + model costs | ★ 50k | 5mo ago | Python | OpenAI API · Claude API |
| CLIO Command Line Intelligence Orchestrator | 69 · Some gaps | CLIFree + model costs | ★ 232 | today | Perl | OpenAI API · Claude API |
| Autohand Code CLI | 62 · Some gaps | CLIFree + model costs | ★ 216 | 2d ago | TypeScript | OpenAI API |
| Oh My Pi | 54 · Major gaps | CLIFree + model costs | ★ 38k | today | TypeScript | Codex · OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows: README mentions automatic git commits, implying write access to repo, but least privilege not explicit; analytics module exists but user confirmation not clear; data flow transparency insufficient; sensitive data handling not explicit; dependencies pinned but no security audit mentioned; external effects (auto-commit) not clearly user-confirmed; git rollback capability exists but not explicit; source attribution not clear. Deductions: lack of explicit permission control, user confirmation, data flow transparency, and sensitive data handling.
Evidence shows: README and pyproject.toml descriptions consistent, good self-consistency; dependency list complete but availability not verified; failure messages not documented. Deduction: insufficient failure message documentation.
Evidence shows: README clearly targets developers, scenarios diverse; capability boundaries described (supports many LLMs, languages); trigger method clear (CLI); environment fit documented (Python versions, Docker). Deductions: capability boundaries and trigger precision not detailed enough.
Evidence shows: README well-structured with install, usage, FAQ; install notes present; naming stable (aider); examples and FAQ present; known limitations not explicit; license Apache-2.0; versioning and changelog mentioned; maintenance responsibility via GitHub repo. Deduction: known limitations not explicitly listed.
Evidence shows: output usability described (auto-commit, lint/test); marginal value supported by user testimonials; cost-benefit mentioned (free open source). Deduction: lack of concrete data.
Evidence shows: README claims have links, but no verifiable test results; cross-source corroboration insufficient; facts and inferences not clearly separated. Deductions: lack of verifiable test evidence and cross-source validation.
- Automatic git commits may overwrite user changes; use with caution.
- Analytics module may collect usage data; users should be aware and control it.
- Many dependencies; regular security audits needed.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
v0.86.0 Aider v0.86.0
Release notes excerpt
- Added support for all GPT-5 models. - Added support for Grok-4 via xai/grok-4 and openrouter/x-ai/grok-4 model names. - Added support for gemini/gemini-2.5-flash-lite-preview-06-17 model, by Tamir Zahavi-Brunner. - /clear now prints “All chat history cleared.” so you know it worked, by Zexin Yuan. - /undo output now shows only the first line of each commit message, making it easier to read. - Added support for openrouter/moonshotai/kimi-k2 model, by Jack Harrington. - Display model announcements with no-arg /model command. …View on GitHub ↗ -
v0.85.0 Aider v0.85.0
Release notes excerpt
New Models - Added support for new Gemini models including gemini-2.5-pro, gemini-2.5-flash, and gemini-2.5-pro-preview-06-05 with thinking tokens support. - Updated model aliases: flash now points to gemini-2.5-flash and gemini now points to gemini-2.5-pro. - Support for Responses API models like o1-pro, o3-pro. - Added support for OpenAI o3-pro model across multiple providers. - Enhanced thinking tokens support: can now be disabled by setting to 0, and improved help text with examples. - Increased max tokens for Deepseek models to 65536 for better performance. …View on GitHub ↗ -
v0.84.0 Aider v0.84.0
Release notes excerpt
Model support & provider updates - Added support for new Claude models including the Sonnet 4 and Opus 4 series (e.g., claude-sonnet-4-20250514, claude-opus-4-20250514) across various providers. The default sonnet and opus aliases were updated to these newer versions. - Added support for the vertex_ai/gemini-2.5-flash-preview-05-20 model. - Updated default OpenRouter models during onboarding to deepseek/deepseek-r1:free for the free tier and anthropic/claude-sonnet-4 for paid tiers. - Automatically refresh GitHub Copilot tokens when used as OpenAI API keys, by Lih Chen. …View on GitHub ↗
FAQ
Am I locked into one model provider?
What credentials do cloud models need?
deepseek=<key>, anthropic=<key>, and openai=<key>.