✨ Code AI
Code AI is an AI-powered CLI agent for data analysis that interacts with Agent Runtimes via the AG-UI and ACP protocols.
Built on Pydantic AI, it provides an interactive terminal experience for code analysis, Jupyter notebooks, and data science workflows — all from your terminal.
Quick Start
pip install codeai
codeai
This starts an interactive session with a background codeai server and the default agent spec. See the CLI page for all options.
Features
- Interactive TUX — Claude Code-inspired terminal interface with Rich rendering, slash commands, and keyboard shortcuts
- Agent Specs — launch preconfigured agents from the agentspecs catalogue with
--agentspec-idor pick one interactively - MCP Tools — agents can use MCP servers (search, file access, APIs) as tools
- Codemode — MCP tools are exposed as programmatic tools the LLM can compose into code
- Streaming — real-time response streaming with tool call visualization
- AG-UI & ACP — connect to local or remote agent servers via HTTP/SSE or WebSocket
- Context Tracking — visualize token usage, export conversation context
Documentation
📄️ Code AI
Code AI is an AI-powered CLI agent for data analysis that interacts with Agent Runtimes via the AG-UI and ACP protocols.
📄️ CLI
Code AI provides a command-line interface built with Typer. Run codeai --help for the full option list.
📄️ Implementation
Code AI is built on Pydantic AI and the Agent Runtimes platform. This page covers the architecture, key components, and design decisions.