mods
An archived AI CLI for shell pipelines on macOS, Linux and Windows. Connects to LocalAI or cloud APIs and saves conversations locally under an MIT license.
mods is a command-line AI tool for people who want to ask questions about command output or use model responses in shell pipelines. It connects to local LLMs through LocalAI as well as cloud services. The project is archived and no longer maintained.
Its main use is working with text already in the terminal. It can combine command output with a prompt, accept a prompt on its own, and request responses in Markdown, JSON or other text formats. That makes it useful for developers and shell users who want AI responses alongside their existing commands rather than in a separate chat window.
The client runs on macOS, Linux and Windows, with FreeBSD support through ports. Model processing happens at the endpoint you choose: LocalAI runs models locally, while OpenAI, Azure OpenAI, Cohere, Groq and Google Gemini receive input through their cloud APIs. Those cloud connections require API keys. mods also works with OpenAI-compatible endpoints, and its documented LocalAI support includes GPT4ALL-J.
Conversations stay local by default. You can return to a saved conversation, continue a previous exchange or turn off conversation saving. Custom roles let you reuse system prompts for recurring tasks, such as requesting shell commands with minimal explanation. MCP support lets it connect to servers that expose tools to the model.
mods is open source under the MIT license. Charm's Crush includes much of its functionality in a non-interactive mode.
