tlm

tlm is a local CLI copilot powered by Ollama: a terminal companion that suggests commands, explains them, and answers questions using open-source language models running entirely on your own workstation. It has about 1,500 stars on GitHub, was actively updated in 2026, and needs no API key, no subscription, and no internet connection.

Overview

Cloud copilots are convenient until you are offline, on a restricted network, or pasting sensitive commands into someone else’s service. tlm takes the opposite approach — everything runs locally through Ollama, so your shell history and questions never leave the machine. It automatically detects your shell (Bash, Zsh, or PowerShell) and helps in the moments terminal users actually struggle: generating a one-liner from a plain description, explaining what an unfamiliar command does before you run it, and answering questions with a lightweight retrieval step that uses your context. You can experiment freely with local models such as Llama 3.3, Phi-4, DeepSeek-R1, and Qwen 2.5 Coder, tuning parameters to trade speed against quality on your hardware.

Key Features

• Command suggestions and one-liner generation from plain descriptions
• Plain-language explanations of commands before you run them
• Ask questions with one-liner RAG over your context
• Fully local via Ollama — no API key, no internet, no subscription
• Automatic shell detection for Bash, Zsh, and PowerShell
• Works on macOS, Linux, and Windows
• Swap local models (Llama, Phi, DeepSeek-R1, Qwen) with your own parameters

💳 Pricing

tlm is free and open source under the Apache-2.0 license. There is nothing to subscribe to: install it, pull a local model with Ollama, and the only resource it consumes is your own machine’s compute.

Who Is It For

Developers, sysadmins, and command-line learners who want copilot-style help in the terminal while keeping everything private and offline — and anyone curious about how far today’s open-source local models can go as everyday shell assistants.