{"id":167356,"date":"2026-05-03T10:30:13","date_gmt":"2026-05-03T07:30:13","guid":{"rendered":"https:\/\/computingforgeeks.com\/aider-cheat-sheet\/"},"modified":"2026-05-04T15:36:30","modified_gmt":"2026-05-04T12:36:30","slug":"aider-cheat-sheet","status":"publish","type":"post","link":"https:\/\/computingforgeeks.com\/aider-cheat-sheet\/","title":{"rendered":"Aider Cheat Sheet – AI Pair Programming Commands and Workflows"},"content":{"rendered":"\n

Aider is a git-native AI pair programming tool that runs in your terminal, edits your files, and writes a commit per change. The CLI exposes 175+ flags, four chat modes, 43 in-chat slash commands, and a YAML config schema with around 100 keys. The official docs split that surface across many pages. This cheat sheet condenses everything into one reference, with every command tested on a real macOS install of Aider 0.86.2 against multiple LLM providers.<\/p>\n\n\n\n

If you came from the Aider setup guide<\/a> and want one page open while you work, this is it.<\/p>\n\n\n\n

Verified working: May 2026 with Aider 0.86.2, tested on macOS against Claude Haiku 4.5 and Sonnet 4.6 (via OpenRouter), with real cost and token counts captured from a fresh test repository.<\/em><\/p>\n\n\n\n

Quick install reminder<\/h2>\n\n\n\n

The recommended path on Linux and macOS is uv tool install<\/code>, which pins a working Python and pulls Aider into an isolated environment:<\/p>\n\n\n\n

uv tool install aider-chat\naider --version<\/code><\/pre>\n\n\n\n
aider 0.86.2<\/code><\/pre>\n\n\n\n
The pipx install aider-install<\/code> bootstrapper still works but adds a layer. uv<\/code> is faster and avoids Python version mismatches. For provider keys, .env layout, and Windows install paths see the dedicated Aider AI pair programming setup<\/a>.<\/p>\n\n\n\n
Where Aider stores config<\/h2>\n\n\n\n
Aider reads four sources in priority order. Top of the list wins:<\/p>\n\n\n\n
    \n\n
  1. CLI flags: aider --model openrouter\/anthropic\/claude-sonnet-4.6<\/code><\/li>\n\n\n
  2. Environment variables: every flag has an AIDER_*<\/code> twin (AIDER_MODEL<\/code>, AIDER_OPENAI_API_KEY<\/code>, AIDER_AUTO_COMMITS<\/code>)<\/li>\n\n\n
  3. Config file: .aider.conf.yml<\/code> searched in git root, then current working directory, then $HOME<\/code><\/li>\n\n\n
  4. Built-in defaults<\/li>\n\n<\/ol>\n\n\n\n
    API keys belong in a separate .env<\/code> file (default location: project root, also gitignored automatically by the first-run prompt):<\/p>\n\n\n\n
    # .env (in your repo root)\nANTHROPIC_API_KEY=sk-ant-api03-...\nOPENAI_API_KEY=sk-proj-...\nGEMINI_API_KEY=aiza...\nOPENROUTER_API_KEY=sk-or-v1-...\nDEEPSEEK_API_KEY=sk-...<\/code><\/pre>\n\n\n\n
    And a working .aider.conf.yml<\/code> validated end to end on macOS:<\/p>\n\n\n\n
    # .aider.conf.yml\nmodel: openrouter\/anthropic\/claude-sonnet-4.6\nweak-model: openrouter\/anthropic\/claude-haiku-4.5\neditor-model: openrouter\/anthropic\/claude-sonnet-4.6\nauto-commits: true\nauto-lint: true\nno-show-model-warnings: true\nno-show-release-notes: true\nlint-cmd:\n  - \"python: ruff check --fix\"\n  - \"javascript: eslint --fix\"\n  - \"go: golangci-lint run --fix\"\ntest-cmd: \"pytest -x\"\nsuggest-shell-commands: true\nnotifications: true\nshow-diffs: true\nchat-history-file: .aider.chat.history.md\ninput-history-file: .aider.input.history\nalias:\n  - \"haiku:openrouter\/anthropic\/claude-haiku-4.5\"\n  - \"sonnet:openrouter\/anthropic\/claude-sonnet-4.6\"\n  - \"opus:openrouter\/anthropic\/claude-opus-4.7\"<\/code><\/pre>\n\n\n\n
    First-run behaviour and auto-gitignore<\/h2>\n\n\n\n
    On the first launch in a git repo, Aider asks to add its history files to .gitignore<\/code>. Real prompt:<\/p>\n\n\n\n
    Add .aider* to .gitignore (recommended)? (Y)es\/(N)o [Yes]: Y\nAdded .aider* to .gitignore<\/code><\/pre>\n\n\n\n
    Skip this with --no-gitignore<\/code>. Always answer yes unless you have a strong reason. .aider.chat.history.md<\/code> contains complete prompts and responses, including any code or configuration the model handled.<\/p>\n\n\n\n
    Top-level CLI flags reference<\/h2>\n\n\n\n
    The full flag list is around 175 entries. Run aider --shell-completions bash<\/code> for the exhaustive set. The table below is the working reference for the flags you actually use day to day.<\/p>\n\n\n\n
    \n
    Flag<\/th>Env var twin<\/th>Purpose<\/th><\/tr><\/thead>
    --model MODEL<\/code><\/td>AIDER_MODEL<\/code><\/td>Main chat model. Format: provider\/model-id<\/code>.<\/td><\/tr>\n
    --weak-model MODEL<\/code><\/td>AIDER_WEAK_MODEL<\/code><\/td>Cheap model for commit messages and summaries.<\/td><\/tr>\n
    --editor-model MODEL<\/code><\/td>AIDER_EDITOR_MODEL<\/code><\/td>Editor model used in architect mode.<\/td><\/tr>\n
    --editor-edit-format FMT<\/code><\/td>AIDER_EDITOR_EDIT_FORMAT<\/code><\/td>Edit format for the editor model.<\/td><\/tr>\n
    --list-models QUERY<\/code><\/td>AIDER_LIST_MODELS<\/code><\/td>Print every model whose name contains QUERY.<\/td><\/tr>\n
    --alias \"name:model\"<\/code><\/td>AIDER_ALIAS<\/code><\/td>Define a model alias. Repeatable.<\/td><\/tr>\n
    --api-key PROVIDER=KEY<\/code><\/td>AIDER_API_KEY<\/code><\/td>Set provider key inline.<\/td><\/tr>\n
    --set-env VAR=VAL<\/code><\/td>AIDER_SET_ENV<\/code><\/td>Set arbitrary env vars before starting.<\/td><\/tr>\n
    --reasoning-effort LEVEL<\/code><\/td>AIDER_REASONING_EFFORT<\/code><\/td>Reasoning knob for o-series and Sonnet 4.6+ thinking.<\/td><\/tr>\n
    --thinking-tokens N<\/code><\/td>AIDER_THINKING_TOKENS<\/code><\/td>Cap on thinking tokens.<\/td><\/tr>\n
    --edit-format FMT<\/code><\/td>AIDER_EDIT_FORMAT<\/code><\/td>diff<\/code>, whole<\/code>, udiff<\/code>, diff-fenced<\/code>.<\/td><\/tr>\n
    --architect<\/code><\/td>AIDER_ARCHITECT<\/code><\/td>Switch on architect mode (planner + editor).<\/td><\/tr>\n
    --auto-accept-architect<\/code><\/td>AIDER_AUTO_ACCEPT_ARCHITECT<\/code><\/td>Skip the architect\/editor confirmation. Default true.<\/td><\/tr>\n
    --map-tokens N<\/code><\/td>AIDER_MAP_TOKENS<\/code><\/td>Repo map token budget. 0<\/code> disables.<\/td><\/tr>\n
    --map-refresh MODE<\/code><\/td>AIDER_MAP_REFRESH<\/code><\/td>auto<\/code> (default), always<\/code>, files<\/code>, manual<\/code>.<\/td><\/tr>\n
    --cache-prompts<\/code><\/td>AIDER_CACHE_PROMPTS<\/code><\/td>Use prompt caching where the provider supports it.<\/td><\/tr>\n
    --cache-keepalive-pings N<\/code><\/td>AIDER_CACHE_KEEPALIVE_PINGS<\/code><\/td>Ping every 5 minutes for N intervals to keep the cache warm.<\/td><\/tr>\n
    --max-chat-history-tokens N<\/code><\/td>AIDER_MAX_CHAT_HISTORY_TOKENS<\/code><\/td>Compact older history above this threshold.<\/td><\/tr>\n
    --no-git<\/code><\/td>AIDER_GIT<\/code><\/td>Run without git (no auto-commits, no diff flow).<\/td><\/tr>\n
    --no-auto-commits<\/code><\/td>AIDER_AUTO_COMMITS<\/code><\/td>Keep git but disable per-edit commits.<\/td><\/tr>\n
    --dirty-commits<\/code> \/ --no-dirty-commits<\/code><\/td>AIDER_DIRTY_COMMITS<\/code><\/td>Commit pre-existing dirty changes when Aider runs.<\/td><\/tr>\n
    --commit<\/code><\/td>AIDER_COMMIT<\/code><\/td>Commit existing changes with an LLM-written message, then exit.<\/td><\/tr>\n
    --commit-prompt PROMPT<\/code><\/td>AIDER_COMMIT_PROMPT<\/code><\/td>Override the system prompt used to write commit messages.<\/td><\/tr>\n
    --dry-run<\/code><\/td>AIDER_DRY_RUN<\/code><\/td>Show edits, do not write files. Verified.<\/td><\/tr>\n
    --lint<\/code><\/td>AIDER_LINT<\/code><\/td>Lint and fix existing code, then exit.<\/td><\/tr>\n
    --lint-cmd \"lang: cmd\"<\/code><\/td>AIDER_LINT_CMD<\/code><\/td>Per-language lint command. Repeatable.<\/td><\/tr>\n
    --auto-lint<\/code><\/td>AIDER_AUTO_LINT<\/code><\/td>Lint after every edit cycle. Default on.<\/td><\/tr>\n
    --test<\/code><\/td>AIDER_TEST<\/code><\/td>Run tests and let Aider iterate fixes.<\/td><\/tr>\n
    --test-cmd CMD<\/code><\/td>AIDER_TEST_CMD<\/code><\/td>Test command Aider uses for --test<\/code> and --auto-test<\/code>.<\/td><\/tr>\n
    --auto-test<\/code><\/td>AIDER_AUTO_TEST<\/code><\/td>Run tests after every successful edit.<\/td><\/tr>\n
    --watch-files<\/code><\/td>AIDER_WATCH_FILES<\/code><\/td>Re-read changed files automatically. Hands-free AI!<\/code> markers.<\/td><\/tr>\n
    --message MSG<\/code> \/ -m<\/code><\/td>AIDER_MESSAGE<\/code><\/td>Send single message, exit.<\/td><\/tr>\n
    --message-file PATH<\/code> \/ -f<\/code><\/td>AIDER_MESSAGE_FILE<\/code><\/td>Read the one-shot prompt from a file.<\/td><\/tr>\n
    --load PATH<\/code><\/td>AIDER_LOAD<\/code><\/td>Replay a saved \/command<\/code> file on launch.<\/td><\/tr>\n
    --apply FILE<\/code><\/td>AIDER_APPLY<\/code><\/td>Apply a SEARCH\/REPLACE diff file Aider produced earlier. Verified.<\/td><\/tr>\n
    --apply-clipboard-edits<\/code><\/td>AIDER_APPLY_CLIPBOARD_EDITS<\/code><\/td>Apply edits pasted into the clipboard.<\/td><\/tr>\n
    --show-repo-map<\/code><\/td>AIDER_SHOW_REPO_MAP<\/code><\/td>Print the repo map and exit. Useful for debugging.<\/td><\/tr>\n
    --show-prompts<\/code><\/td>AIDER_SHOW_PROMPTS<\/code><\/td>Print the assembled system prompts and exit.<\/td><\/tr>\n
    --browser<\/code> \/ --gui<\/code><\/td>AIDER_GUI<\/code><\/td>Launch the Streamlit browser UI.<\/td><\/tr>\n
    --copy-paste<\/code><\/td>AIDER_COPY_PASTE<\/code><\/td>Clipboard ferry mode for offline workflows.<\/td><\/tr>\n
    --vim<\/code><\/td>AIDER_VIM<\/code><\/td>VI editing mode in the input.<\/td><\/tr>\n
    --multiline<\/code><\/td>AIDER_MULTILINE<\/code><\/td>Multi-line input (Meta-Enter submits).<\/td><\/tr>\n
    --yes-always<\/code><\/td>AIDER_YES_ALWAYS<\/code><\/td>Auto-confirm every prompt.<\/td><\/tr>\n
    -c, --config PATH<\/code><\/td>AIDER_CONFIG<\/code><\/td>Use an explicit config file.<\/td><\/tr>\n
    --env-file PATH<\/code><\/td>AIDER_ENV_FILE<\/code><\/td>Use an explicit .env file.<\/td><\/tr>\n
    --upgrade<\/code><\/td>AIDER_UPGRADE<\/code><\/td>Upgrade in place. Verified.<\/td><\/tr>\n
    --just-check-update<\/code><\/td>AIDER_JUST_CHECK_UPDATE<\/code><\/td>Exit code reflects whether a new version is available.<\/td><\/tr>\n
    --shell-completions SHELL<\/code><\/td>AIDER_SHELL_COMPLETIONS<\/code><\/td>Print bash, zsh, or tcsh completions.<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n
    Model selection and provider routing<\/h2>\n\n\n\n
    Aider speaks every major LLM provider through LiteLLM. The model string is provider\/model-id<\/code>. Skip the provider for OpenAI:<\/p>\n\n\n\n
    # Anthropic (direct)\naider --model anthropic\/claude-sonnet-4-6\naider --model anthropic\/claude-opus-4-7\n\n# OpenAI (no provider prefix)\naider --model gpt-5\naider --model o3\n\n# Google\naider --model gemini\/gemini-2.5-pro\naider --model gemini\/gemini-2.5-flash\n\n# DeepSeek\naider --model deepseek\/deepseek-chat\naider --model deepseek\/deepseek-reasoner\n\n# AWS Bedrock (requires AWS_PROFILE)\naider --model bedrock\/anthropic.claude-sonnet-4-6\n\n# OpenRouter (single key, all providers, validated working in this guide)\naider --model openrouter\/anthropic\/claude-sonnet-4.6\naider --model openrouter\/anthropic\/claude-haiku-4.5\naider --model openrouter\/google\/gemini-2.5-pro\n\n# Local Ollama\naider --model ollama_chat\/llama3.3:70b\naider --model ollama_chat\/qwen2.5-coder:32b<\/code><\/pre>\n\n\n\n
    If you misspell the model name, Aider suggests close matches. Real warning captured during testing:<\/p>\n\n\n\n
    Warning for openrouter\/anthropic\/claude-haiku-4-5: Unknown context window size and costs.\nDid you mean one of these?\n- openrouter\/anthropic\/claude-3-5-haiku\n- openrouter\/anthropic\/claude-3-haiku\n- openrouter\/anthropic\/claude-haiku-4.5<\/code><\/pre>\n\n\n\n
    Filter the full model list from the CLI:<\/p>\n\n\n\n
    $ aider --list-models claude\nModels which match \"claude\":\n- anthropic\/claude-haiku-4-5\n- anthropic\/claude-opus-4-7\n- anthropic\/claude-sonnet-4-6\n- anthropic\/claude-sonnet-4-5-20250929\n- ... 90+ more across providers<\/code><\/pre>\n\n\n\n
    Built-in aliases<\/h3>\n\n\n\n
    Aider ships preset shortcuts for the common models. Use them anywhere a model name would go:<\/p>\n\n\n\n
    \n
    Alias<\/th>Resolves to<\/th><\/tr><\/thead>
    opus<\/code><\/td>claude-opus-4-7<\/code><\/td><\/tr>\n
    sonnet<\/code><\/td>claude-sonnet-4-6<\/code><\/td><\/tr>\n
    haiku<\/code><\/td>claude-haiku-4-5<\/code><\/td><\/tr>\n
    4o<\/code><\/td>gpt-4o<\/code><\/td><\/tr>\n
    mini<\/code><\/td>gpt-4o-mini<\/code><\/td><\/tr>\n
    flash<\/code><\/td>gemini-2.5-flash<\/code><\/td><\/tr>\n
    deepseek<\/code><\/td>deepseek\/deepseek-chat<\/code><\/td><\/tr>\n
    r1<\/code><\/td>deepseek\/deepseek-reasoner<\/code><\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n
    Override or extend with --alias<\/code> on the CLI or under the alias:<\/code> key in .aider.conf.yml<\/code>. CLI definitions win, then config-file aliases, then built-in aliases. Once defined, switch mid-session with \/model haiku<\/code>.<\/p>\n\n\n\n
    Main, weak, and editor models<\/h3>\n\n\n\n