terminal-aichat

README中文

A CLI for AI/LLM chat in terminal

• written in rust, light (6.5MB binary size), super fast.
• multi platform(Windows, Linux, MacOS)
• using /v1/chat/completion API

aichat <INPUT MESSAGE>

Quick Start

Installation

sh

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/slow-groovin/terminal-aichat/releases/latest/download/terminal-aichat-installer.sh | sh

cargo

cargo install terminal-aichat

homebrew

brew install slow-groovin/tap/terminal-aichat

npm

Deprecated, but you can still install version 0.3.5 with it

npm install terminal-aichat@latest

powershell

powershell -ExecutionPolicy Bypass -c "irm https://github.com/slow-groovin/terminal-aichat/releases/latest/download/terminal-aichat-installer.ps1 | iex"

binary

or download executable binaries directly in Release page.

Prerequisites

Edit the config file to setup providers and models:

# First, get the config file location
aichat --list
# Or just:
aichat

# Then edit the config file (example for Linux)
nano ~/.config/terminal-aichat/config.jsonc

The config file contains a commented-out example that you can uncomment and modify.

Chat

# Directly send a message
aichat how to view ubuntu release version

# If your message conflicts with an option, wrap it with quotes
aichat "how to use --config option"

# other ways
aichat "<INPUT MESSAGE>"
aichat -- <INPUT MESSAGE>

# pipe
cat input.txt | aichat
cat input.txt | aichat "explain this"

# pure mode (display for model/prompts configs and costs will be hide)
aichat --pure "Hello?"

Configuration

View Configurations

aichat --list
# or
aichat -l

This will show all configured providers, models, prompts, and the config file location.

Edit Configuration

Configuration is done by directly editing the config file (JSONC format with comments support).

Config file locations (cross-platform):

• Linux: ~/.config/terminal-aichat/config.jsonc
• macOS: ~/Library/Application Support/terminal-aichat/config.jsonc
• Windows: %APPDATA%\terminal-aichat\config.jsonc

The config file uses JSONC format, which supports comments. A complete example is provided in the default config file.

Example config structure:

{
  "providers": {
    "openai": {
      "name": "OpenAI",
      "baseURL": "https://api.openai.com/v1",
      "apiKey": "sk-...",
      "models": {
        "gpt-4o": {
          "name": "gpt-4o",
          "temperature": 0.7
        },
        "gpt-5-mini": {
          "name": "gpt-5-mini",
          "temperature": 0.5
        }
      }
    },
    "openrouter": {
      "name": "OpenRouter",
      "baseURL": "https://openrouter.ai/api/v1",
      "apiKey": "sk-or-...",
      "models": {
        "meta-llama/llama-3-70b-instruct": {
          "name": "llama-3",
          "temperature": 0.3
        }
      }
    }
  },
  "prompts": {
    "sample_prompt": {
      "content": "You are a terminal assistant. You are giving help to user in the terminal. Give concise responses whenever possible. Because of terminal cannot render markdown, DO NOT contain any markdown syntax(`,```, #, ...) in your response, use plain text only.\n"
    }
  },
  "default-model": "gpt-5-mini",
  "default-prompt": "sample_prompt",
  "disable-stream": false,
  "pure": false,
  "verbose": false
}

Specify a Model

# Use a specific model selector (searches all providers)
# - model key: "gpt-4o"
# - or provider/model_key: "openrouter/meta-llama/llama-3-70b-instruct"
aichat --model gpt-4o "Hello?"

# Or short form
aichat -m gpt-5-mini "Hello?"

# When model keys conflict across providers, or when model keys contain '/',
# use provider/model_key to disambiguate:
aichat -m openrouter/meta-llama/llama-3-70b-instruct "Hello?"

Specify a Prompt

# Use a specific prompt
aichat --prompt concise "Hello?"

# Or short form
aichat -p sample_prompt "Hello?"

Use Custom Config File

# Specify a custom config file path
aichat --config /path/to/config.jsonc --list
aichat --config /path/to/config.jsonc "Hello?"

Use Temporary API Key via Environment Variable

Useful for avoiding persistent API key storage or for testing. It will override API key in final request.

export OPENAI_API_KEY=sk-***************
aichat "Hello?"

Set Log Level

export LOG_LEVEL=DEBUG

Equivalent to using --verbose

Pure Mode (--pure)

Suppresses all extra messages and outputs only the response.

aichat --pure "Hello?"

Verbose Logging (--verbose)

aichat --verbose "Hello?"

Disable Streaming Mode (--disable-stream)

aichat --disable-stream "Hello?"