What is q?

q is a small command line tool, written in C, that blends normal shell commands with questions to an AI language model in the same terminal prompt. The idea is that instead of switching to a browser whenever you get stuck on a command, you can ask your question right there and get an answer, similar to having an experienced colleague sitting next to you. Running q in its interactive REPL mode gives you a numbered prompt where you can type ordinary shell requests in plain English, such as showing the last 100 nginx log lines, restarting a service, or finding large files modified recently. The tool sends that text to a local, OpenAI-compatible language model server and streams the response back to the terminal as it arrives. If the model's answer includes example shell commands in code blocks, those get numbered automatically so you can run one just by typing its number and a period, without retyping it. By default q talks to a language model server running on your own machine at 127.0.0.1, though the server address and port can be changed with environment variables, and an API key can be set if the server requires one. It also supports connecting to external MCP servers over HTTP, which extend the set of tools the model can call beyond the built in ones for reading time, reading files, and writing files. Other features include saving and resuming past sessions, keeping conversation context across multiple questions if you choose to, logging API requests and responses to a file, and a set of REPL slash commands for adjusting settings like timeouts and turn limits while the tool is running. The author describes q as a rough proof of concept built with around seven thousand lines of AI generated code, used daily as a personal tool, and released so others can experiment with it rather than waiting for a polished version. It requires libcurl development headers to build and installs with a standard make and make install workflow.