tlm
A local AI terminal assistant that generates and explains shell commands through Ollama. Runs offline on macOS, Linux and Windows under Apache 2.0.
tlm is an open source terminal assistant for people who want help writing shell commands and understanding unfamiliar ones. It runs models on your workstation through Ollama, so command assistance doesn't depend on a cloud service. It works offline and requires no API key or subscription.
You describe a task in plain language, and tlm suggests a one-line command. You can also give it a command to explain, which is useful when checking shell syntax or trying to understand what a copied command does. It detects Bash, Zsh and PowerShell automatically, adapting its assistance to your shell on macOS, Linux or Windows.
Model choice is flexible. Supported choices include Llama 3.3, Phi4, DeepSeek-R1 and Qwen, with Qwen2.5-Coder as the default. You can choose a different model for a particular request and adjust the response style. This makes it an option for people who already use Ollama and want shell assistance with models they run locally.
The assistant keeps command prompts and model processing on your workstation. tlm is written in Go and uses the Apache 2.0 license.
Its beta ask command can use files from a chosen directory as context, with include and exclude patterns for local question answering.
