aica - AI Code Analyzer
@dotneet
aica(AI Code Analyzer) reviews your code using AI. Supports CLI and GitHub Actions.
概览
What is aica - AI Code Analyzer?
aica - AI Code Analyzer is an open-source, customizable, platform-independent AI coding agent and code review tool. It supports the Model Context Protocol (MCP) with stdio and SSE transports, integrates with GitHub, and can be used as a CLI or MCP server.
How to use aica - AI Code Analyzer?
Install via Bun, clone the repository, run bun install && bun run build, and place the binary in your PATH. Set a GITHUB_TOKEN and at least one LLM provider (Anthropic, OpenAI, or Gemini) via environment variables or a aica.toml configuration file. Use commands like aica review, aica agent, aica mcp server, or aica commit.
Key features of aica - AI Code Analyzer
- AI code review and agent capabilities
- MCP support via stdio and SSE transports
- Automatic knowledge retrieval for code review
- Symbol- and vector-based code/document search
- Generate summaries, commit messages, and pull requests
- Slack notification and GitHub Actions integration
Use cases of aica - AI Code Analyzer
- Automate code reviews on local diffs or GitHub pull requests
- Execute AI agents to perform file-based tasks or answer prompts
- Generate commit messages and pull request descriptions
- Integrate as an MCP server with AI assistants
- Send code review notifications to Slack
FAQ from aica - AI Code Analyzer
How does aica differ from pr-agent or Cursor?
aica is open source, platform-independent, and does not rely on GitHub or other hosting services; it can be fully customized and used locally.
What are the runtime dependencies?
You need Bun to build the binary, a GitHub token (GITHUB_TOKEN or gh auth token), and an API key for at least one LLM provider (Anthropic, OpenAI, or Gemini).
Where does aica store its data?
aica creates index databases for code and documents during aica reindex. The README does not specify the exact storage location, but data is stored locally in the repository or configuration directory.
What are known limitations of the agent command?
The agent command has the potential to break your file system. It is recommended to use Anthropic Claude 3.5 Sonnet for agent tasks.
What transports and authentication does the MCP server support?
The MCP server supports stdio and SSE transports. Authentication uses the configured GitHub token and LLM API keys from environment variables or aica.toml.
