Mods was a command-line tool that let you send text to an AI language model directly from your terminal, making it useful inside shell scripts and pipelines. The core idea was simple: pipe any text into Mods, add a question or instruction as an argument, and it would forward everything to an AI model and print the response. This made it easy to do things like summarize command output, reformat data, or ask questions about text without leaving the terminal. The tool supported several AI providers, including OpenAI, Cohere, Groq, Azure OpenAI, and locally running models via LocalAI. You could choose which model to use via a flag, or be prompted to pick one interactively. Conversations were saved locally by default, each with a short identifier and a title similar to how Git tracks commits. You could list saved conversations, continue a previous one, or delete old ones. For repeated use cases, you could define custom roles in a settings file. A role is a named system prompt that shapes how the AI responds. For example, a shell role might instruct the model to only output single-line commands with no explanation. The tool also had support for Model Context Protocol servers, which is a way of connecting AI tools to external data sources or capabilities. As of March 2026, the project has been archived and is no longer maintained. The team behind it has moved focus to a different tool called Crush, which includes a non-interactive mode called crush run that covers much of what Mods did. The repository remains public and the code is open source for anyone who wants to fork it.
← charmbracelet on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.