From install to chat,in four steps.
The LLMTune CLI brings every model, agent, skill, and prompt to your terminal. Install it, log in, and start composing — no infrastructure to manage.
- Step 01 · Install
Install the CLI
One command installs the native binary to ~/.llmtune/bin (macOS/Linux) or %USERPROFILE%\.llmtune\bin (Windows). No Node required.
zsh$curl -fsSL https://llmtune.io/install.sh | bashInstalled: ~/.llmtune/bin/llmtune$llmtune --versionllmtune/X.XX-Beta - Step 02 · Authenticate
Log in with your API key
Run llmtune login and paste a key. Your token is stored locally in ~/.llmtune/config.json — never uploaded.
zsh$llmtune login→Paste your API key: llmt_••••••••••••••••Saved to ~/.llmtune/config.jsonDon't have a key yet?
- 1. Open the API Keys page in your dashboard.
- 2. Click New key, name it, and copy the
llmt_…value. - 3. Paste it into the prompt — or skip the prompt entirely:
Create an API keynon-interactive$llmtune login --key llmt_your_key_here - Step 03 · Chat
Start a conversation
Launch an interactive session, or one-shot a task with -p. Everything runs against your selected model, streaming straight to the terminal.
zsh$llmtune chatGLM-5.2, streaming. Type /help for commands.>Refactor the auth flow in this reporeading 4 files, working locally - Step 04 · Compose
Pick your model, agent, skill & prompt
Inside chat, tap the arrow keys to swap any part of your stack on the fly — or set them up front with flags. Four primitives, infinite combinations.
zsh↵llmtune chat --model glm-5.2 --agent codepress ← model → agent ↓ skill ↑ prompt↓skill: review-codeequipped review-code, ready←Model
glm-5.2
→Agent
code
↓Skill
review-code
↑Prompt
/diff
You're set. Start building.
Jump into a session from the terminal, or pick up right here in the browser — your models, agents, and skills are the same everywhere.