About aica - AI Code Analyzer
aica is an open-source code analysis CLI for developers, with AI review, agent tasks, and GitHub Actions support. It can also run as an MCP server and retrieve code context through symbol and vector searches.
Key features
- 01AI code review
- 02AI agent tasks
- 03Symbol and vector search
- 04Change summaries and commit messages
- 05GitHub Actions integration
- 06Customizable prompts
Use cases
- 01Review diffs or selected files
- 02Run automated reviews in GitHub Actions
- 03Generate commit messages and pull requests
MCP support
aica can start an MCP server over stdio or SSE. It can also connect to configured MCP servers using an MCP setup file.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEaica - AI Code Analyzer
Motivation
There are already excellent code review tools such as pr-agent and cursor. However, pr-agent relies on code hosting services like GitHub, which limits its usability, and cursor is not open source, meaning it cannot be fully customized or integrated with other tools.
So, I decided to create a new tool with the following characteristics:
- Customizable
- Open Source
- Platform Independent
Features
- AI Coding Agent
- AI Code Review
- MCP(Model Context Protocol) supported. stdio and SSE transports are supported.
- Automatic knowledge retrieving for code review
- Symbol based code search for retrieving knowledge
- Vector based document search for retrieving knowledge
- Generate summary of changes
- Generate commit message
- Create pull request with AI-generated title and body
- Prompt customization
- Slack notification
- Single binary executable by
bun build --compile - GitHub Actions integration. (See wiki page to setup actions.)
Install
Build and install a binary:
# Install bun before build aica.
#
# Official Install Document:
# https://bun.sh/docs/installation#installing
git clone https://github.com/dotneet/aica.git
cd aica
bun install
bun run build
cp ./dist/aica path-to-your-bin-directory
Setup GitHub Token:
# if GITHUB_TOKEN is not set, aica try to get token from `gh auth token`.
export GITHUB_TOKEN=your_github_token
Setup LLM:
# You can set the following items in your environment variables or aica.toml file.
# Configure at least one of the following providers:
# Anthropic (If you use agent, strongly recommend to use claude sonnet model)
export AICA_LLM_PROVIDER=anthropic
export ANTHROPIC_API_KEY=your_api_key
export ANTHROPIC_MODEL=claude-3-5-sonnet-20241022
# OpenAI
export AICA_LLM_PROVIDER=openai
export OPENAI_API_KEY=your_api_key
export OPENAI_MODEL=o3-mini
# Gemini
export AICA_LLM_PROVIDER=google
export GOOGLE_API_KEY=your_api_key
export GOOGLE_MODEL=gemini-2.0-flash
Configuration
You can customize the configuration by creating a aica.toml file.
See aica.example.toml.
aica.toml must be placed in one of the following directories.
- root directory of the repository
- ${HOME}/.config/aica/aica.toml
- ${GITHUB_WORKSPACE}/aica.toml
Language
You can specify the language for AI output. The language can be specified either through AICA_LANGUAGE or in the language section of aica.toml. By default, the language is automatically detected from the LANG environment variable.
export AICA_LANGUAGE=Japanese
Priority:
AICA_LANGUAGElanguage.languagein aica.toml- Automatic detection from
LANG
Usage
Review
# review the diff from HEAD
aica review [options] [pattern]
# review specific files
aica review src/main.ts
# review the files matching the specific glob pattern
aica review "src/**/*.ts"
Options:
--dir: Target directory path--slack: Send notification to Slack
Agent
# execute AI agent with a prompt
aica agent "your prompt here"
# execute AI agent with a instruction file
aica agent -f instruction.txt
# conversation mode. You can interact with AI agent by typing your prompt.
aica agent
This command executes a task using an AI agent. The agent automatically determines and executes the necessary actions based on the given prompt.
Recommend to use anthropic claude 3.5 sonnet for agent.
NOTE: This command has potential to break your file system. Please be careful.
Aica Agent as an MCP Server
# start a MCP server
aica mcp server
# start a MCP server with a config file
aica mcp server --config aica.toml
Reindex
# reindex the code and document databases
aica reindex
Summary
# generate a summary of the diff from HEAD
aica summary [options]
Options:
--dir: Target directory path
Commit Changes
This command commits changes with an AI-generated commit message.
# commit all changes(including untracked and unstaged changes) with an AI-generated commit message
aica commit [options] # commits all changes (including untracked and unstaged changes) with an AI-generated commit message
aica commit --staged # commits only staged changes with an AI-generated commit message
aica commit --push # commits all changes and pushes to remote repository
Options:
--staged: commit only staged changes.--dryRun: Show result without execution--push: Push to remote repository after committing
Create Pull Request
This command creates a pull request on GitHub.
# Creates a pull request. If there are changes, they will be committed automatically.
aica create-pr [options]
# Creates a pull request with only staged changes
aica create-pr --staged
Options:
--withSummary: Generate summary of diff from HEAD (default: true)--body: Pull request body--dryRun: Show result without execution--staged: Only include staged changes
Generate Commit Message
# generate a one-line commit message based on the diff from HEAD
aica commit-message [options]
Options:
--dir: Target directory path
Show Configuration
# show current configuration
aica show-config [options]
Options:
--default: Show default configuration
Other Commands
aica --version: Show version informationaica --help [command]: Show help information for a specific command or general help
Add Context
.cursorrules and .clinerules are automatically added to the context. To customize the context, configure the [rules] section in aica.toml.
Additionally, .cursor/rules/*.mdc files are supported by default. This function can be disabled through the settings.
MCP(Model Context Protocol)
create a mcp.json file to define the MCP server.
example:
[
{
"name": "example-server",
"type": "stdio",
"command": "node",
"args": ["./server.js"]
},
{
"name": "example-server",
"type": "sse",
"url": "http://localhost:3001/sse"
}
]
set setupFile in aica.toml like below.
[mcp]
setupFile = "mcp.json"
GitHub Actions
Setup
See wiki page for details.
Commands
You can interact with AI features by leaving comments with the following commands.
# edit the code with AI
/aica edit "your prompt here"
# update summary in the pull request body
/aica summary
# review the latest diff
/aica review
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.