Skip to main content

Aider AI LLCAI Coding Agent

Aider

Open-source command-line pair programmer, licensed Apache 2.0, that edits the files you add to the chat with help from a map of the whole repository. It commits each change to git and connects to almost any LLM through your own key, including local models.

The 30-Second Take

Pick Aider only at a pinned version, because its latest release shipped in February 2026.

Main has had no commits since May, so the version you pin is the version you support. It is a free terminal pair programmer that edits the files you add to the chat, on almost any model through your own key.

Best fit
Engineers who steer changes file by file in the terminal, on models the team chooses.
Main constraint
Releases stopped at 0.86.2 on February 12, 2026, and commits to main stopped three months later.
First move
Pin one version, enforce git hooks with --git-commit-verify, and publish the models engineers may use.

Why

The trade is control: an engineer picks the model and decides exactly which files it sees. Every dollar of spend goes to that model, and local models work too.

Practitioner sentiment, all practitioners+28 Positive

57 practitioners, 106 sources reviewed, Updated Nov 1, 2026How this is measured

Read the GILD Review
Save a product your organization uses.Keep up with Aider without adding it to your stack.Visit Website
Operating stack
Agentic Software Development

Comparisons

How does Aider compare with the alternatives?

  1. Aider vs Claude Code

    • AI Coding Agents

    Choose Claude Code when the agent should search the repository and run commands on its own, on Claude plans from US$20 a month. Choose Aider when engineers steer each edit on models they pick, with no seat cost.

  2. Aider vs Cline

    • AI Coding Agents

    Choose Cline when engineers want an agent in their IDE that edits and runs commands with approval. Choose Aider when they work in the terminal and want each edit committed to git. Both are free under Apache 2.0 on your own model keys.

Operator Insights

What are engineering leaders learning from deploying Aider?

Buying and deployment

What engineering leaders should verify before a broader rollout.

Sourced from Aider documentation

How you buy it

Free, Apache 2.0

No paid plan, seats, or support contract. Your model provider bills the tokens.

Source 5, pypi.org, aider-chat 0.86.2 on PyPI (JSON API): aider-chat 0.86.2 on PyPI (JSON API)Source 1, Aider AI LLC, Installation: InstallationSource 6, Aider AI LLC, aider.chat home: aider.chat home

Details

Free and open source under Apache 2.0, with no paid plan, seats, or support contract. Install with aider-install, uv, pipx, or a one-line script; model usage is billed by your provider, or runs on a local model.

Cost drivers

Model tokens only

Aider shows the cost of each message and session, and a cheaper model writes commits.

Source 7, aider.chat, In-chat commands: In-chat commandsSource 2, Aider AI LLC, Options reference: Options reference

Details

Model tokens only. The /tokens command reports the tokens the current chat context uses, and a cheaper weak model writes commit messages and chat summaries.

Where it runs

Python CLI

In the terminal on macOS, Linux, and Windows. Editors connect through watched file comments.

Source 1, Aider AI LLC, Installation: InstallationSource 2, Aider AI LLC, Options reference: Options reference

Details

A Python CLI in the terminal on macOS, Linux, and Windows. Editors connect through --watch-files comments and scripts through --message.

What to verify

Auto-commits on

Commits skip pre-commit hooks unless you set --git-commit-verify. Last release February 12, 2026.

Source 2, Aider AI LLC, Options reference: Options referenceSource 8, Aider AI LLC, YAML config file: YAML config fileSource 5, pypi.org, aider-chat 0.86.2 on PyPI (JSON API): aider-chat 0.86.2 on PyPI (JSON API)

Details

Auto-commits are on, and commits skip pre-commit hooks unless --git-commit-verify is set. A .aider.conf.yml at the repository root loads automatically. The latest release is 0.86.2, shipped February 12, 2026.

Signals

What is the current evidence saying about Aider?

A current read of what developers, teams, and technical leaders report publicly about Aider, with every finding linked to its sources.

Whose sources count

Practitioner sentiment, all practitioners

+28 Positive net

of 59 sources that take a side, each author counted once.

How this figure is made

19 more state no overall view and cast no vote.

Leaders and teams have 3 accounts; GILD prints their sentiment from 20.

Current sentiment

Positive, mixed and negative sources

The score counts them as 28 positive, 17 mixed and 12 negative votes, one per author.

Across public accounts the read is positive: model choice and explicit context control are the most praised capabilities. Practitioners split almost evenly on whether Aider's developer-driven control beats agentic tools that explore and run commands on their own, and the slower release pace is the newest concern.

Use a task pilot to measure accepted work and review time.

Evidence field. All 106 sources, one mark each.

Arrange by

Each mark is one source: positive, mixed, negative, no opinion stated, or documentation (not counted).

By sentiment, All practitioners. Every source GILD read, grouped as the field draws it.
GroupSourcesPositiveMixedNegativeNo opinion stated and not counted in the scoreDocumentation, checked and not counted
Positive30300000
Mixed17017000
Negative12001200
Not counted26000260
Documentation21000021
Every source GILD read, by the quarter it was published, as the field's time axis counts it. Documentation is dated by the day it was checked.
QuarterSources
Q1 202519
Q2 202517
Q3 202514
Q4 202511
Q1 202613
Q2 20265
Q3 20265
Q4 20261
Documentation, checked21

By sentiment

21 documentation pages check facts and are not counted.

Hover a mark for a quick look, click it or press Enter to open its source.Tap a mark to open its source, then step to its neighbors with Previous and Next.

What practitioners praise

Top praise of 6. Each mark is a source on it; filled marks praise it.

  1. Any Model, Any ProviderAll 15 sources on it praise it

Most common concerns

Top concern of 5. Each mark is a source on it; filled marks raise it.

  1. Slower Release Pace10 of 11 sources on it raise it as a concern
All 106 sources behind these findings
  1. Aider home pageAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  2. InstallationAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  3. Connecting to LLMsAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  4. GitHub Copilot as a providerAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  5. Git integrationAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  6. Chat modesAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  7. Repository mapAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  8. TipsAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  9. Specifying coding conventionsAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  10. Aider in your IDEAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  11. Linting and testingAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  12. Options referenceAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  13. YAML config fileAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  14. AnalyticsAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  15. Scripting aiderAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  16. LLM leaderboardsAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  17. Release historyAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  18. FAQAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  19. Aider on GitHubAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  20. Aider releases on GitHubAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  21. aider-chat on PyPIAider AI LLC (Official), Product documentation, checked Oct 12, 2026
  22. Comment on: Thoughts and questions about the future direction of aiderAltchemist, Junior to intermediate software engineer (self-described: Recently adopted after avoiding AI tools), GitHub issue comment, published Oct 8, 2026
  23. Comment on: Anomalyco/OpenCode, Aider-AI/aider, and Cline Reviewparetolaw, Hacker News commenter, self-described, Hacker News comment, published Aug 27, 2026
  24. AI Coding Benchmark: Claude Code vs CursorSedat Dogan and Şevval Alper, AIMultiple (self-described: Independent benchmark: 10 full-stack web tasks run one-shot, CLI agents on Claude Sonnet 4.6), Independent benchmark, AIMultiple, published Aug 21, 2026
  25. Comment on: Aider exits with code 0 on fatal API Connection errors, breaking headless automation pipelineschernistry, Maintainer of an orchestrator that runs CLI coding agents (self-described: Orchestrator running coding agents in parallel git worktrees, with Aider as an adapter), GitHub issue comment, published Aug 12, 2026
  26. Aider exits with code 0 on fatal API Connection errors, breaking headless automation pipelinesnoelsaw1, Engineer running Aider in headless automation (self-described: Headless automation loop), GitHub issue, published Aug 11, 2026
  27. Aider makes changes in ask modegberkolaiko, Aider user filing a GitHub issue (self-described: Editing a LaTeX file), GitHub issue, published Jul 1, 2026
  28. Opening an untrusted repository with aider executes arbitrary commands at startup via .aider.conf.yml test-cmd / lint-cmd (no confirmation)geo-chen, Security researcher filing a GitHub issue (self-described: Reproduced on current main), GitHub issue, published Jun 12, 2026
  29. I Tried Building a Complex Security Tool with a 1.5B Local Model: Here's What BrokeMusa Nayyer, Independent developer (self-described: Personal forensic data pipeline project on Lubuntu), Personal blog post, DEV Community, published May 27, 2026
  30. I stopped forcing every coding job through Claude and started using this insteadAnurag Singh, Technology journalist at XDA Developers (self-described: Personal coding work), Independent review, XDA Developers, published May 22, 2026
  31. Aider Architect Mode for Service Scaffolding: Output Quality and Required Follow-UpRaghuveer, Software engineer (self-described: Scaffolding a Go, gRPC, Postgres and Redis microservice from a specification), Personal blog post, iamraghuveer.com, published Apr 25, 2026
  32. Comment on: I don't want your PRs anymoredpc_01234, Hacker News commenter, self-described, Hacker News comment, published Apr 22, 2026
  33. Comment on: Ask HN: How is AI-assisted coding going for you professionally?KronisLV, Hacker News commenter (self-described: Anthropic Max subscription with Claude Code), Hacker News comment, published Mar 16, 2026
  34. Comment on: Beating Aider's 20% pass rate on local Qwen 32B using deterministic RAGebercruzdev, Hacker News commenter (self-described: Local qwen2.5-coder 32B on Ollama), Hacker News comment, published Mar 16, 2026
  35. When AI Coding Tools Break Your UI: A Lesson From Aider, Claude & Geminiramamurthy valavandan, Developer building a personal publishing app (self-described: Next.js application, AgentPost), Personal blog post, DEV Community, published Mar 12, 2026
  36. Comment on: No release notes for v0.86.2Magneticdud, Aider user commenting on a GitHub issue (self-described: Aider 0.86.2 with gpt-5.3-codex), GitHub issue comment, published Mar 3, 2026
  37. Comment on: Thoughts and questions about the future direction of aiderroebi, Aider user commenting on a GitHub issue (self-described: Open source developer in test automation), GitHub issue comment, published Feb 24, 2026
  38. Comment on: Writing code is cheap nowchr15m, Freelance full-stack developer (self-described: Self-managed freelancer), Hacker News comment, published Feb 24, 2026
  39. Comment on: I was banned from Claude for scaffolding a Claude.md file?thunfischtoast, Hacker News commenter (self-described: Anthropic models through OpenRouter, compared with a Claude Code subscription), Hacker News comment, published Jan 23, 2026
  40. Comment on: Ask HN: Do you have any evidence that agentic coding works?mythrwy, Hacker News commenter (self-described: Compared with Codex CLI), Hacker News comment, published Jan 21, 2026
  41. Comment on: Thoughts and questions about the future direction of aiderfingon, Aider user commenting on a GitHub issue (self-described: Work), GitHub issue comment, published Jan 12, 2026
  42. Comment on: The next two years of software engineeringnoufalibrahim, Hacker News commenter (self-described: Personal site-blocker project of about 4,000 lines), Hacker News comment, published Jan 12, 2026
  43. Comment on: Anthropic blocks third-party use of Claude Code subscriptionsaserafini, Hacker News commenter, self-described, Hacker News comment, published Jan 9, 2026
  44. Thoughts and questions about the future direction of aideryiouyou, Aider user filing a GitHub issue (self-described: Daily development workflow), GitHub issue, published Jan 7, 2026
  45. Comment on: Vibe Coding Killed Cursornoo_u, Hacker News commenter (self-described: Config files and small utility scripts), Hacker News comment, published Jan 2, 2026
  46. Comment on: Ask HN: How can I get better at using AI for programming?devdp430, Hacker News commenter (self-described: Go codebase of thousands of files), Hacker News comment, published Dec 14, 2025
  47. Comment on: Mistral releases Devstral2 and Mistral Vibe CLItroyvit, Hacker News commenter (self-described: Watch mode from the editor), Hacker News comment, published Dec 9, 2025
  48. Comment on: Mistral releases Devstral2 and Mistral Vibe CLIzmmmmm, Hacker News commenter, self-described, Hacker News comment, published Dec 9, 2025
  49. Comment on: Gemini CLI tips and tricks for agentic codingtherealmarv, Hacker News commenter, self-described, Hacker News comment, published Nov 27, 2025
  50. Background Coding Agents: Context Engineering (Honk, Part 2)Max Charas, Senior Staff Engineer, and Marc Bruggmann, Principal Engineer, at Spotify (Spotify Fleet Management; automated code migrations across thousands of repositories), Company engineering blog post, Spotify Engineering, published Nov 24, 2025
  51. Comment on: The new Aider-CE fork of Aider is now officialaitchnyu, Hacker News commenter (self-described: Edits by leaving AI comments in the IDE), Hacker News comment, published Nov 17, 2025
  52. What is the intended future of Aider?adrianwebb, Aider user filing a GitHub issue, self-described, GitHub issue, published Nov 15, 2025
  53. Comment on: Ask HN: Who uses open LLMs and coding assistants locally? Share setup and laptoplreeves, Hacker News commenter (self-described: Local gpt-oss-120b, compared with hosted GPT-5 and Claude), Hacker News comment, published Oct 31, 2025
  54. Where is Paul?ErichBSchulz, Aider user (self-described: Daily Aider user following the project's community), GitHub issue on Aider-AI/aider, published Oct 30, 2025
  55. Comment on: No new releases since early AugKreijstal, Aider user commenting on a GitHub issue, self-described, GitHub issue comment, published Oct 14, 2025
  56. No new releases since early Augwhoabuddy, Aider user filing a GitHub issue (self-described: Upgraded Aider before every session), GitHub issue, published Oct 13, 2025
  57. Comment on: Claude Code 2.0flyinglizard, Hacker News commenter, self-described, Hacker News comment, published Sep 29, 2025
  58. Comment on: Development speed is not a bottlenecktharkun__, Hacker News commenter (self-described: 15-year-old monorepo of about 100 services), Hacker News comment, published Sep 9, 2025
  59. What I'm up to (September 2025)Aaron Sumner, Software Engineering Manager at O'Reilly Media (Team maintaining nearly fifty software systems and libraries, some with long-dead test suites and no documentation), Personal blog post, aaronsumner.com, published Sep 7, 2025
  60. Comment on: I'm absolutely rightdingnuts, Hacker News commenter, self-described, Hacker News comment, published Sep 5, 2025
  61. Comment on: AWS CEO says using AI to replace junior staff is 'Dumbest thing I've ever heard'JeremyNT, Hacker News commenter, self-described, Hacker News comment, published Aug 21, 2025
  62. Comment on: Claude Sonnet 4 now supports 1M tokens of contextfaangguyindia, Hacker News commenter (self-described: DeepSeek R1 architect with a Qwen3 Coder editor), Hacker News comment, published Aug 12, 2025
  63. Comment on: I watched Gemini CLI hallucinate and delete my filesmelodyogonna, Hacker News commenter (self-described: Python project with mypy, compared with Gemini CLI), Hacker News comment, published Jul 23, 2025
  64. Comment on: My experience with Claude Code after two weeks of adventuressingron, Hacker News commenter (self-described: Godot project in GDScript), Hacker News comment, published Jul 18, 2025
  65. Comment on: LLM Inevitabilismcess11, Hacker News commenter (self-described: Background in enterprise Java at a government agency), Hacker News comment, published Jul 15, 2025
  66. Comment on: Kiro: A new agentic IDENitpickLawyer, Hacker News commenter, self-described, Hacker News comment, published Jul 14, 2025
  67. Comment on: Opencode: AI coding agent, built for the terminalgwd, Hacker News commenter (self-described: Compared with opencode), Hacker News comment, published Jul 7, 2025
  68. Comment on: Thesis: Interesting work is less amenable to the use of AIjavcasas, Hacker News commenter (self-described: 200-line Python simulator on a local Ollama model), Hacker News comment, published Jul 7, 2025
  69. Bug report - aider disgards changes if the LM makes changes *and* adds files (which user agrees to) in same response.ErichBSchulz, Aider user (self-described: Personal Python project, plan file edited in diff mode), GitHub issue on Aider-AI/aider, published Jul 5, 2025
  70. Comment on: Claude Code now supports hooksstavros, Hacker News commenter, self-described, Hacker News comment, published Jul 1, 2025
  71. Comment on: Claude Code feels like magic because it is iterativeanonzzzies, Hacker News commenter speaking for a team (self-described: Small team building parsers and compilers), Hacker News comment, published Jun 18, 2025
  72. Comment on: Andrej Karpathy's talk on the future of the industrymsgodel, Hacker News commenter (self-described: Local Qwen3 8B through llama.cpp), Hacker News comment, published Jun 18, 2025
  73. Hello Team, how can we deploy AIDER (currently CLI) like a plugin in VSCode, Visual Studio Professional and IntelliJ.rohitttttt, Developer packaging Aider for colleagues (self-described: Company team on VS Code, Visual Studio and IntelliJ), GitHub issue, published Jun 12, 2025
  74. Comment on: Agentic Coding RecommendationsBeetleB, Hacker News commenter (self-described: Gemini for personal work; Copilot at work), Hacker News comment, published Jun 12, 2025
  75. Comment on: Agentic Coding RecommendationsKarrot_Kream, Hacker News commenter (self-described: Mostly Aider, compared with Codex), Hacker News comment, published Jun 12, 2025
  76. Comment on: Ask HN: Anyone struggling to get value out of coding LLMs?bird0861, Hacker News commenter (self-described: Hosted models and local models on a personal RTX 4090), Hacker News comment, published Jun 10, 2025
  77. Comment on: Cursor 1.0fragmede, Hacker News commenter (self-described: Parallel Aider copies in git worktrees), Hacker News comment, published Jun 5, 2025
  78. AI Assistants and uv: Which Dev Tools Deliver?Calvin Hendryx-Parker, CTO at Six Feet Up (Six Feet Up, a Python and AI consultancy; live IndyPy comparison of five assistants converting a Django project's GraphQL API to REST), Company news post, Six Feet Up, published May 28, 2025
  79. Comment on: Ask HN: Anyone struggling to get value out of coding LLMs?morsecodist, Hacker News commenter (self-described: Next.js app), Hacker News comment, published May 26, 2025
  80. Comment on: The unreasonable effectiveness of an LLM agent loop with tool useTeMPOraL, Hacker News commenter (self-described: A CONVENTIONS.md in each project), Hacker News comment, published May 16, 2025
  81. Comment on: AI Is Like a Crappy Consultantgawa, Hacker News commenter, self-described, Hacker News comment, published May 13, 2025
  82. Comment on: Ask HN: Cursor or Windsurf?m3adow, Hacker News commenter (self-described: Small to mid-scope changes), Hacker News comment, published May 12, 2025
  83. Comment on: A flat pricing subscription for Claude Codevidarh, Hacker News commenter (self-described: Runs an AI, devops and cloud cost consultancy), Hacker News comment, published May 9, 2025
  84. Comment on: Ask HN: How much better are AI IDEs vs. copy pasting into chat apps?Ey7NFZ3P0nzAe, Hacker News commenter, self-described, Hacker News comment, published May 8, 2025
  85. Basic Claude CodeHarper Reed, CEO of 2389 Research (self-described: Spec-driven side projects; previously an Aider workflow), Personal blog post, harper.blog, published May 8, 2025
  86. Feature: Get Aider to implement features from jira ticketsLuisGoCity, Engineer on a company ML team (self-described: Company ML team using Aider daily), GitHub issue, published Apr 23, 2025
  87. Comment on: Firebase Studiopiokoch, Hacker News commenter, self-described, Hacker News comment, published Apr 10, 2025
  88. Git integration is all-or-noneujisati, Aider user filing a GitHub issue (self-described: Work repository), GitHub issue, published Mar 28, 2025
  89. Aider using no-verify by default for commitsaskkhan84, Aider user filing a GitHub issue for a team (self-described: Team repositories with git hooks), GitHub issue, published Mar 20, 2025
  90. Claude Code vs AiderAndrew Keenan Richardson, Independent developer (self-described: Command-line AI coding on a personal code base), Personal blog post, Mechanistic Mind (Substack), published Mar 20, 2025
  91. Comment on: I've been using Claude Code for a couple of daysjamil7, Hacker News commenter (self-described: Swift codebase with dependencies in other repositories), Hacker News comment, published Mar 11, 2025
  92. Comment on: Llama.cpp AI Performance with the GeForce RTX 5090 Reviewregularfry, Hacker News commenter (self-described: Local Ollama models in architect mode), Hacker News comment, published Mar 10, 2025
  93. Comment on: I've been using Claude Code for a couple of daysinciampati, Hacker News commenter (self-described: Switches between model providers), Hacker News comment, published Mar 10, 2025
  94. Comment on: I've been using Claude Code for a couple of daysBeetleB, Hacker News commenter, self-described, Hacker News comment, published Mar 9, 2025
  95. Comment on: Ask HN: Is Cursor deleting working code for you too or is it just me?i_love_retros, Hacker News commenter (self-described: At work, where DeepSeek's API is not allowed), Hacker News comment, published Mar 8, 2025
  96. Comment on: Aider: Using Uv as an Installerzitterbewegung, Hacker News commenter, self-described, Hacker News comment, published Mar 6, 2025
  97. Comment on: Claude 3.7 Sonnet and Claude Codedavely, Hacker News commenter (self-described: Legacy React refactor at work), Hacker News comment, published Feb 24, 2025
  98. Comment on: Claude 3.7 Sonnet and Claude Codegwd, Hacker News commenter (self-described: Daily use on a Go codebase), Hacker News comment, published Feb 24, 2025
  99. My LLM codegen workflow atmHarper Reed, CEO of 2389 Research (self-described: Greenfield and legacy side projects), Personal blog post, harper.blog, published Feb 16, 2025
  100. Comment on: Zed now predicts your next edit with Zeta, our new open modeltomw1808, Hacker News commenter (self-described: Aider in a Docker container in the IDE terminal), Hacker News comment, published Feb 14, 2025
  101. 100% cpu freezing does not respond to ctrl c on latest releasetrashhalo, Aider user filing a GitHub issue (self-described: Markdown rendering in the terminal), GitHub issue, published Feb 9, 2025
  102. Comment on: Promising results from DeepSeek R1 for codejvanderbot, Hacker News commenter (self-described: Rust Bevy project), Hacker News comment, published Jan 28, 2025
  103. Comment on: When AI promises speed but delivers debugging helljoshstrange, Hacker News commenter, self-described, Hacker News comment, published Jan 26, 2025
  104. Comment on: Aider in Your IDEretrodaredevil, Hacker News commenter (self-described: IntelliJ terminal, local Llama 3.1 8B and GPT-4o), Hacker News comment, published Jan 12, 2025
  105. Comment on: Aider in Your IDEhboon, Hacker News commenter (self-described: Neovim), Hacker News comment, published Jan 10, 2025
  106. Comment on: How I program with LLMsryanobjc, Hacker News commenter (self-described: Emacs), Hacker News comment, published Jan 7, 2025

Jan 2025 to Oct 2026.

85 sources from practitioners, by month and sentiment.

The full readExplore all Aider Signals6 praise themes, 6 concerns, 4 points of agreement and 3 disagreements across all three groups, with all 106 sources open to inspect.

Where It Fits

Where does Aider belong in the stack?

Open Where It Fits: the Stack Field and the full map

Aider is Core across Build, supports Plan, Validate, and Ship, and touches Operate.

Business area
Product & Engineering
Function
Software Development
Operating stack
Agentic Software Development
Primary role
Terminal Pair Programmer
Aider beside the rest of its stackThe Stack Field, every product on every layer
FootprintAider across the Agentic Software Development stack. Bar height is its role on each layer.

  1. Plan

    1. Work planning: Supports

      Work planning, supports. Ask mode discusses a change without editing files and architect mode has one model plan the edits another model makes, but backlog and task planning stay in the team's planning tool.

  2. Build

    1. Codebase context: Core

      Codebase context, core. Aider sends a tree-sitter map of the whole git repository, with its key classes, functions, and signatures, alongside every request, and the developer chooses which files the model can edit.

    2. Coding: Core

      Coding, core. This is Aider's primary role: it writes and edits code across the files in the chat, in most popular languages, with the model and edit format the developer selects.

    3. Agent runtime: Core

      Agent runtime, core. Aider runs as an interactive chat in the terminal with in-chat commands, acts on AI comments written in any editor with --watch-files, and runs single instructions from scripts with --message.

    4. Tool execution: Supports

      Tool execution, supports. Aider runs shell commands it suggests after confirmation and adds their output to the chat with /run; it has no built-in MCP client, so tool integrations stay with other agents.

  3. Validate

    1. Debugging: Supports

      Debugging, supports. Aider can read command and test output added with /run or /test and fix the errors it shows, while investigation across logs and production data stays with debugging tools.

    2. Testing: Supports

      Testing, supports. Aider lints every file it edits by default, runs a test command with /test or after each change with --auto-test, and tries to fix failures. The test suite and CI remain the team's.

    3. Code review: Touches

      Code review, touches. Aider's per-edit commits and /diff give reviewers a clean history, but it does not review pull requests.

    4. Application security: Touches

      Application security, touches. Aider is not a security scanner; teams set its confirmation and config-file defaults as part of their own security policy.

  4. Ship

    1. Source control: Supports

      Source control, supports. Aider commits each change with a generated message, commits preexisting edits separately, and offers /undo, /diff, and /commit, while branching, pull requests, and the repository host stay with the team's source control.

    2. CI/CD: Touches

      CI/CD, touches. Aider can be scripted into automation with --message and --yes-always, but it does not build, deploy, or run pipelines.

  5. Operate

    1. Observability: Touches

      Observability, touches. Aider can work on fixes when given error output, but it does not monitor or alert on production systems.

Market mapThe layers Aider covers, and who else owns each one.

  1. Codebase context & intelligence

    CoreBuild

  2. Coding & implementation

    CoreBuild

  3. Agent interface & developer runtime

    CoreBuild

  4. Work planning & task decomposition

    SupportsPlan

  5. Tool execution & integrations

    SupportsBuild

7 more layers on the full map. Core, Supports and Touches describe coverage. They say nothing about quality.

Models

Which models run in Aider?

The current releases GILD tracks that public sources record running in Aider. Each model's page charts its public benchmark scores.

  1. gpt-oss-120bOpenAI's largest open-weight model, 117B parameters with 5.1B active, with a 131,072-token context window. Weights under the Apache 2.0 license; OpenAI lists no API price for it, so it runs on your own hardware or a host's.gpt-ossBenchmarks

More on Aider

  1. GILD ReviewGILD's evaluation, with the buying facts
  2. Research2 pieces
  3. Guides4 guides
  4. Operator Insights1 operator account
  5. Where It FitsOn the Agentic Software Development stack
  6. Related Products8 products