What it is

term-llm

term-llm is a terminal-first AI runtime.

It turns natural language into command suggestions, supports persistent chat, runs agents, edits files, calls tools, schedules jobs, and works with local or hosted models.

If you are new, install it, configure one provider, run the quickstart, and then come back for the workflow or reference page that matches the job.

Start with the quickstart Open configuration reference

$ term-llm exec "find the 3 biggest files"

$ term-llm ask @reviewer "review the staged changes"

$ term-llm chat @codebase

Install

Copy the command and run it

curl -fsSL https://raw.githubusercontent.com/samsaffron/term-llm/main/install.sh | sh

The full command is shown inline, so you can see exactly what you are about to paste.

Start with intent

Three common entry points

Exec

`term-llm exec`

Use plain English to get a command suggestion fast. Good for one-shot terminal tasks like finding the 3 biggest files.

Ask + agents

`term-llm ask @reviewer`

Use a built-in agent when the job has a mode. @reviewer is read-only and git-aware, so it reviews code without wandering off into edits.

Chat + agents

`term-llm chat @codebase`

Start a persistent session when you need back-and-forth. @codebase is the right tool for tracing behavior and understanding a repository.

Read the right page

Pick the lane that matches the job

Start here

New here

Quickstart for the shortest path from install to a working first run
Installation if you just want the install command and platform notes
Providers and setup to wire up OpenAI, Anthropic, Gemini, Ollama, OpenRouter, and the rest
Usage guide when you want the mental model for exec, ask, and chat

Do a thing

Common workflows

Skills for portable instruction bundles that add task-specific context
Agents for built-in agents and when to use them
File editing for safe edit workflows from the terminal
MCP servers to connect external tools and services
Job runner for scheduled and background automation

Exact details

Reference shortcuts

Configuration reference for config keys, file locations, and examples
Providers and models for provider-specific settings and model naming
Built-in tools for tool behavior and availability
Sessions for persistence, resume behavior, and storage details

Model underneath

Architecture and internals

Architecture overview for how runtimes, tools, routing, skills, and memory fit together
Memory guide for persistent memory behavior and fragment management
Web UI and API for browser access and HTTP endpoints
Debugging when something feels off and you need the shortest path to evidence