The CLI by Supermodeltools is an MCP server tool built with Go, designed to connect AI clients with Supermodel graph tools. It processes codebases, uploads them to the Supermodel API, and builds code graphs. This program helps expose Supermodel's capabilities to AI agents via the Model Context Protocol. It generates .graph files that AI agents read to understand code structure, helping to manage token usage. Distributed as a binary or npm package, CLI supports developers working with various AI agents.

What does CLI connect to?

The CLI connects to the Supermodel API to upload code repositories and build detailed code graphs. It functions as an MCP server tool, making Supermodel's graph analysis available to AI clients through the Model Context Protocol. While the CLI itself generates plain text .graph files for agents to read via file system operations, the Model Context Protocol generally supports stdio, SSE, or streamable HTTP for AI client communication. Compatible AI clients include Claude Code, Cursor, Copilot, Windsurf, Aider, and any agent able to read files.

Core features and tools

The CLI exposes specific callable tools for developers. The main supermodel command maps codebases, uploads them to the Supermodel API, builds call graphs, and writes .graph files. These files are plain text, read by AI agents using file system commands like "grep" and "cat". Additionally, supermodel skill appends an awareness prompt about .graph files to an agent's instructions, teaching agents how to interpret and use these generated files for improved context and decision-making.

Who is it for?

CLI is for developers who work with AI agents and need to provide them with a deep understanding of code structures. It helps agents interpret codebases efficiently by supplying pre-analyzed graph data. This tool is valuable for anyone aiming to enhance their AI agent's ability to navigate and reason about complex software projects. It supports agents like Claude Code, Cursor, Copilot, Windsurf, and Aider. Skip this if your AI agent workflows do not involve reading local files for code context.

How to get started

The server authenticates using an API key. Setup involves running the "supermodel" command, which initiates an automatic setup wizard for authentication and repository confirmation.

Use cases

• Generating detailed code graphs from a codebase, saving them as .graph sidecar files for AI agents to understand structure.
• Maintaining live, updated code graphs by running continuously in the background, ensuring agents always have current code relationship data.
• Augmenting AI agent instructions by injecting prompts that teach agents how to interpret and utilize generated .graph files.

The CLI provides a method for integrating code analysis directly into AI agent workflows. It produces .graph files that agents use to gain insights into code structure, supporting more informed decisions and efficient processing. The tool also helps keep agent context current through live updates. Try generating a code graph for your repository using the "supermodel" command to see how it enhances your AI agent's understanding.