DevCats 8d56ee8182
fix: resolve aider icon path to correct icon (#102)
- code.svg to aider.svg in README.md

---------

Co-authored-by: M Atif Ali <atif@coder.com>
2025-05-15 12:54:01 -05:00
..

display_name description icon maintainer_github verified tags
Aider Run Aider AI pair programming in your workspace ../../../../.icons/aider.svg coder true
agent
ai
aider

Aider

Run Aider AI pair programming in your workspace. This module installs Aider and provides a persistent session using screen or tmux.

module "aider" {
  source   = "registry.coder.com/coder/aider/coder"
  version  = "1.0.1"
  agent_id = coder_agent.example.id
}

Features

  • Interactive Parameter Selection: Choose your AI provider, model, and configuration options when creating the workspace
  • Multiple AI Providers: Supports Anthropic (Claude), OpenAI, DeepSeek, GROQ, and OpenRouter
  • Persistent Sessions: Uses screen (default) or tmux to keep Aider running in the background
  • Optional Dependencies: Install Playwright for web page scraping and PortAudio for voice coding
  • Project Integration: Works with any project directory, including Git repositories
  • Browser UI: Use Aider in your browser with a modern web interface instead of the terminal
  • Non-Interactive Mode: Automatically processes tasks when provided via the task_prompt variable

Module Parameters

Parameter Description Type Default
agent_id The ID of a Coder agent (required) string -
folder The folder to run Aider in string /home/coder
install_aider Whether to install Aider bool true
aider_version The version of Aider to install string "latest"
use_screen Whether to use screen for running Aider in the background bool true
use_tmux Whether to use tmux instead of screen for running Aider in the background bool false
session_name Name for the persistent session (screen or tmux) string "aider"
order Position of the app in the UI presentation number null
icon The icon to use for the app string "/icon/aider.svg"
experiment_report_tasks Whether to enable task reporting bool true
system_prompt System prompt for instructing Aider on task reporting and behavior string See default in code
task_prompt Task prompt to use with Aider string ""
ai_provider AI provider to use with Aider (openai, anthropic, azure, etc.) string "anthropic"
ai_model AI model to use (can use Aider's built-in aliases like "sonnet", "4o") string "sonnet"
ai_api_key API key for the selected AI provider string ""
custom_env_var_name Custom environment variable name when using custom provider string ""
experiment_pre_install_script Custom script to run before installing Aider string null
experiment_post_install_script Custom script to run after installing Aider string null
experiment_additional_extensions Additional extensions configuration in YAML format to append to the config string null

Note

: use_screen and use_tmux cannot both be enabled at the same time. By default, use_screen is set to true and use_tmux is set to false.

Usage Examples

Basic setup with API key

variable "anthropic_api_key" {
  type        = string
  description = "Anthropic API key"
  sensitive   = true
}

module "aider" {
  count      = data.coder_workspace.me.start_count
  source     = "registry.coder.com/coder/aider/coder"
  version    = "1.0.1"
  agent_id   = coder_agent.example.id
  ai_api_key = var.anthropic_api_key
}

This basic setup will:

  • Install Aider in the workspace
  • Create a persistent screen session named "aider"
  • Configure Aider to use Anthropic Claude 3.7 Sonnet model
  • Enable task reporting (configures Aider to report tasks to Coder MCP)

Using OpenAI with tmux

variable "openai_api_key" {
  type        = string
  description = "OpenAI API key"
  sensitive   = true
}

module "aider" {
  count       = data.coder_workspace.me.start_count
  source      = "registry.coder.com/coder/aider/coder"
  version     = "1.0.1"
  agent_id    = coder_agent.example.id
  use_tmux    = true
  ai_provider = "openai"
  ai_model    = "4o" # Uses Aider's built-in alias for gpt-4o
  ai_api_key  = var.openai_api_key
}

Using a custom provider

variable "custom_api_key" {
  type        = string
  description = "Custom provider API key"
  sensitive   = true
}

module "aider" {
  count               = data.coder_workspace.me.start_count
  source              = "registry.coder.com/coder/aider/coder"
  version             = "1.0.1"
  agent_id            = coder_agent.example.id
  ai_provider         = "custom"
  custom_env_var_name = "MY_CUSTOM_API_KEY"
  ai_model            = "custom-model"
  ai_api_key          = var.custom_api_key
}

Adding Custom Extensions (Experimental)

You can extend Aider's capabilities by adding custom extensions:

module "aider" {
  count      = data.coder_workspace.me.start_count
  source     = "registry.coder.com/coder/aider/coder"
  version    = "1.0.1"
  agent_id   = coder_agent.example.id
  ai_api_key = var.anthropic_api_key

  experiment_pre_install_script = <<-EOT
  pip install some-custom-dependency
  EOT

  experiment_additional_extensions = <<-EOT
  custom-extension:
    args: []
    cmd: custom-extension-command
    description: A custom extension for Aider
    enabled: true
    envs: {}
    name: custom-extension
    timeout: 300
    type: stdio
  EOT
}

Note: The indentation in the heredoc is preserved, so you can write the YAML naturally.

Task Reporting (Experimental)

This functionality is in early access as of Coder v2.21 and is still evolving. For now, we recommend testing it in a demo or staging environment, rather than deploying to production

Learn more in the Coder documentation

Join our Discord channel or contact us to get help or share feedback.

Your workspace must have either screen or tmux installed to use this.

Task reporting is enabled by default in this module, allowing you to:

  • Send an initial prompt to Aider during workspace creation
  • Monitor task progress in the Coder UI
  • Use the coder_parameter resource to collect prompts from users

Setting up Task Reporting

To use task reporting effectively:

  1. Add the Coder Login module to your template
  2. Configure the necessary variables to pass the task prompt
  3. Optionally add a coder_parameter to collect prompts from users

Here's a complete example:

module "coder-login" {
  count    = data.coder_workspace.me.start_count
  source   = "registry.coder.com/modules/coder-login/coder"
  version  = "1.0.15"
  agent_id = coder_agent.example.id
}

variable "anthropic_api_key" {
  type        = string
  description = "Anthropic API key"
  sensitive   = true
}

data "coder_parameter" "ai_prompt" {
  type        = "string"
  name        = "AI Prompt"
  default     = ""
  description = "Write a prompt for Aider"
  mutable     = true
  ephemeral   = true
}

module "aider" {
  count       = data.coder_workspace.me.start_count
  source      = "registry.coder.com/coder/aider/coder"
  version     = "1.0.1"
  agent_id    = coder_agent.example.id
  ai_api_key  = var.anthropic_api_key
  task_prompt = data.coder_parameter.ai_prompt.value

  # Optionally customize the system prompt
  system_prompt = <<-EOT
You are a helpful Coding assistant. Aim to autonomously investigate
and solve issues the user gives you and test your work, whenever possible.
Avoid shortcuts like mocking tests. When you get stuck, you can ask the user
but opt for autonomy.
YOU MUST REPORT ALL TASKS TO CODER.
When reporting tasks, you MUST follow these EXACT instructions:
- IMMEDIATELY report status after receiving ANY user message.
- Be granular. If you are investigating with multiple steps, report each step to coder.
Task state MUST be one of the following:
- Use "state": "working" when actively processing WITHOUT needing additional user input.
- Use "state": "complete" only when finished with a task.
- Use "state": "failure" when you need ANY user input, lack sufficient details, or encounter blockers.
Task summaries MUST:
- Include specifics about what you're doing.
- Include clear and actionable steps for the user.
- Be less than 160 characters in length.
  EOT
}

When a task prompt is provided via the task_prompt variable, the module automatically:

  1. Combines the system prompt with the task prompt into a single message in the format:
SYSTEM PROMPT:
[system_prompt content]

This is your current task: [task_prompt]
  1. Executes the task during workspace creation using the --message and --yes-always flags
  2. Logs task output to $HOME/.aider.log for reference

If you want to disable task reporting, set experiment_report_tasks = false in your module configuration.

Using Aider in Your Workspace

After the workspace starts, Aider will be installed and configured according to your parameters. A persistent session will automatically be started during workspace creation.

Session Options

You can run Aider in three different ways:

  1. Direct Mode: Aider starts directly in the specified folder when you click the app button
  • Simple setup without persistent context
  • Suitable for quick coding sessions
  1. Screen Mode (Default): Run Aider in a screen session that persists across connections
  • Session name: "aider" (or configured via session_name)
  1. Tmux Mode: Run Aider in a tmux session instead of screen
  • Set use_tmux = true to enable
  • Session name: "aider" (or configured via session_name)
  • Configures tmux with mouse support for shared sessions

Persistent sessions (screen/tmux) allow you to:

  • Disconnect and reconnect without losing context
  • Run Aider in the background while doing other work
  • Switch between terminal and browser interfaces

Available AI Providers and Models

Aider supports various providers and models, and this module integrates directly with Aider's built-in model aliases:

Provider Example Models/Aliases Default Model
anthropic "sonnet" (Claude 3.7 Sonnet), "opus", "haiku" "sonnet"
openai "4o" (GPT-4o), "4" (GPT-4), "3.5-turbo" "4o"
azure Azure OpenAI models "gpt-4"
google "gemini" (Gemini Pro), "gemini-2.5-pro" "gemini-2.5-pro"
cohere "command-r-plus", etc. "command-r-plus"
mistral "mistral-large-latest" "mistral-large-latest"
ollama "llama3", etc. "llama3"
custom Any model name with custom ENV variable -

For a complete and up-to-date list of supported aliases and models, please refer to the Aider LLM documentation and the Aider LLM Leaderboards which show performance comparisons across different models.

Troubleshooting

If you encounter issues:

  1. Screen/Tmux issues: If you can't reconnect to your session, check if the session exists with screen -list or tmux list-sessions
  2. API key issues: Ensure you've entered the correct API key for your selected provider
  3. Browser mode issues: If the browser interface doesn't open, check that you're accessing it from a machine that can reach your Coder workspace

For more information on using Aider, see the Aider documentation.