You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

15 KiB

AIChat: All-in-one AI CLI Tool

CI Crates Discord

AIChat is an all-in-one AI CLI tool featuring chat REPL, RAG, function calling, and more.

AIChat Command

AIChat Chat-REPL

Key Features

  • Unified LLM Client: Access 100+ LLMs across 20+ AI platforms effortlessly.
  • RAG Retrieval: Get answer enhanced by your documents and knowledge bases.
  • Function Calling: Connect LLMs to external tools seamlessly.
  • Execute Commands: Use natural language to run shell commands.
  • Chat REPL: Powerful and feature-rich interactive chat interface.
  • Useful Roles: Tailor LLM behavior with customizable roles.
  • Unlimited Sessions: Automatic message compression for endless conversations.
  • Shell Auto-Completion: AI-based auto-completion for shell commands.
  • Highly Customizable: Customize configurations and themes to your preferences.
  • Local API Server: Host a local server with OpenAI-compatible API.
  • LLM Playground/Arena: Experiment and compete LLMs in WebUI.

Supported LLMs

  • OpenAI GPT-3.5/GPT-4 (paid, vision, embedding, function-calling)
  • Gemini: Gemini-1.0/Gemini-1.5 (free, paid, vision, embedding, function-calling)
  • Claude: Claude-3 (vision, paid, function-calling)
  • Mistral (paid, embedding, function-calling)
  • Cohere: Command-R/Command-R+ (paid, embedding, function-calling)
  • Reka (paid, vision)
  • Perplexity: Llama-3/Mixtral (paid)
  • Groq: Llama-3/Mixtral/Gemma (free)
  • Ollama (free, local, embedding)
  • Azure OpenAI (paid, vision, embedding, function-calling)
  • VertexAI: Gemini-1.0/Gemini-1.5 (paid, vision, embedding, function-calling)
  • VertexAI-Claude: Claude-3 (paid, vision)
  • Bedrock: Llama-3/Claude-3/Mistral (paid, vision)
  • Cloudflare (free, paid, vision)
  • Replicate (paid)
  • Ernie (paid)
  • Qianwen (paid, vision, embedding)
  • Moonshot (paid)
  • Deepseek (paid)
  • ZhipuAI: GLM-3.5/GLM-4 (paid, vision)
  • LingYiWanWu (paid)
  • Other openAI-compatible platforms


Package Managers

  • Rust Developers: cargo install aichat
  • Homebrew/Linuxbrew Users: brew install aichat
  • Pacman Users: yay -S aichat
  • Windows Scoop Users: scoop install aichat
  • Android Termux Users: pkg install aichat

Pre-built Binaries

Download pre-built binaries for macOS, Linux, and Windows from GitHub Releases, extract them, and add the aichat binary to your $PATH.


Upon first launch, AIChat will guide you through the configuration process.

> No config file, create a new one? Yes
> AI Platform: openai
> API Key: <your_api_key_here>
✨ Saved config file to '<user-config-dir>/aichat/config.yaml'

Feel free to adjust the configuration according to your needs.

💡 Use the AICHAT_CONFIG_DIR environment variable to custom the config dir for aichat files.

model: openai:gpt-3.5-turbo      # Specify the language model to use
temperature: null                # Set default temperature parameter
top_p: null                      # Set default top-p parameter
save: true                       # Indicates whether to persist the message
save_session: null               # Controls the persistence of the session, if null, asking the user
highlight: true                  # Controls syntax highlighting
light_theme: false               # Activates a light color theme when true
wrap: no                         # Controls text wrapping (no, auto, <max-width>)
wrap_code: false                 # Enables or disables wrapping of code blocks
keybindings: emacs               # Choose keybinding style (emacs, vi)
prelude: null                    # Set a default role or session to start with (role:<name>, session:<name>)

# Command that will be used to edit the current line buffer with ctrl+o
# if unset fallback to $EDITOR and $VISUAL
buffer_editor: null

# Compress session when token count reaches or exceeds this threshold (must be at least 1000)
compress_threshold: 4000

# Controls the function calling feature. For setup instructions, visit
function_calling: false

  - type: openai
    api_key: sk-xxx

  - type: openai-compatible
    name: localai
      - name: llama3
        max_input_tokens: 8192


The config.example.yaml file provides a comprehensive list of all configuration options with detailed explanations. You'll find information on each option's meaning, default value, relevant documentation, and customization guidance.


Usage: aichat [OPTIONS] [TEXT]...

  [TEXT]...  Input text

  -m, --model <MODEL>        Select a LLM model
      --prompt <PROMPT>      Use the system prompt
  -r, --role <ROLE>          Select a role
  -s, --session [<SESSION>]  Start or join a session
      --save-session         Forces the session to be saved
      --serve [<ADDRESS>]    Serve the LLM API and WebAPP
  -e, --execute              Execute commands in natural language
  -c, --code                 Output code only
  -f, --file <FILE>          Include files with the message
  -H, --no-highlight         Turn off syntax highlighting
  -S, --no-stream            Turns off stream mode
  -w, --wrap <WRAP>          Control text wrapping (no, auto, <max-width>)
      --light-theme          Use light theme
      --dry-run              Display the message without sending it
      --info                 Display information
      --list-models          List all available chat models
      --list-roles           List all available roles
      --list-sessions        List all available sessions
  -h, --help                 Print help
  -V, --version              Print version

Here are some practical examples:

aichat                                          # Start REPL

aichat -e install nvim                          # Execute
aichat -c fibonacci in js                       # Code

aichat -s                                       # REPL + New session
aichat -s session1                              # REPL + New/Reuse 'session1'

aichat --info                                   # View system info
aichat -r role1 --info                          # View role info
aichat -s session1 --info                       # View session info

cat data.toml | aichat -c to json > data.json   # Pipe stdio/stdout

aichat -f data.toml -c to json > data.json      # Send files

aichat -f a.png -f b.png diff images            # Send images

Shell commands

Simply input what you want to do in natural language, and aichat will prompt and run the command that achieves your intent.

aichat -e <text>...


AIChat is aware of OS and shell you are using, it will provide shell command for specific system you have. For instance, if you ask aichat to update your system, it will return a command based on your OS. Here's an example using macOS:

$ aichat -e update my system
? sudo softwareupdate -i -a

The same prompt, when used on Ubuntu, will generate a different suggestion:

$ aichat -e update my system
? sudo apt update && sudo apt upgrade -y

Shell integration

This is a very handy feature, which allows you to use aichat shell completions directly in your terminal, without the need to type aichat with prompt and arguments. This feature puts aichat completions directly into terminal buffer (input line), allowing for immediate editing of suggested commands.


To install shell integration, go to ./scripts/shell-integration to download the script and source the script in rc file. After that restart your shell. You can invoke the completion with alt+e hotkey.

Generating code

By using the --code or -c parameter, you can specifically request pure code output.


The -c/--code option ensures the extraction of code from Markdown.


AIChat has a powerful Chat REPL.

REPL Features:

  • Tab auto-completion
  • Custom REPL Prompt
  • Emacs/VI keybinding
  • Edit/paste multi-line text
  • Open an editor to edit the current prompt
  • History and Undo

.help - show help message

> .help
.help                    Show this help message
.info                    View system info
.model                   Change the current LLM
.prompt                  Create a temporary role using a prompt
.role                    Switch to a specific role
.info role               View role info
.exit role               Leave the role
.session                 Begin a chat session
.info session            View session info
.save session            Save the chat to file
.clear messages          Erase messages in the current session
.exit session            End the current session
.file                    Include files with the message
.set                     Adjust settings
.copy                    Copy the last response
.exit                    Exit the REPL

Type ::: to start multi-line editing, type ::: to finish it.
Press Ctrl+O to open an editor to edit the input buffer.
Press Ctrl+C to cancel the response, Ctrl+D to exit the REPL

.info - view information

  • .info: View system information
  • .info role: view your current role information.
  • .info session: view your current session information.

.model - change the current LLM

> .model openai:gpt-4
> .model ollama:llama3

Tab autocompletion helps in quickly typing the model names.

.role - switch to a specific role

Select a role:

> .role emoji

Send message with the role:

emoji> hello

Leave current role:

emoji> .exit role

> hello
Hello there! How can I assist you today?

Temporarily use a role without switching to it:

> .role emoji hello


.session - Begin a chat session

By default, aichat behaves in a one-off request/response manner.

You should run aichat with -s/--session or use the .session command to start a session.

> .session

temp) 1 to 5, odd only                                                                    0
1, 3, 5

temp) to 7                                                                        19(0.46%)
1, 3, 5, 7

temp) .exit session                                                               42(1.03%)
? Save session? (y/N)  

.prompt - create a temporary role using a prompt

There are situations where setting a system message is necessary, but modifying the roles.yaml file is undesirable. To address this, we leverage the .prompt to create a temporary role specifically for this purpose.

> .prompt your are a js console


.file - read files and send them as input

Usage: .file <file>... [-- text...]

.file message.txt
.file config.yaml -- convert to toml
.file screentshot.png -- design a web app based on the image
.file -- what is the difference?

The capability to process images through .file command depends on the current models vision support.

.rag - chat with your documents and knowledge bases.

> .rag test1
> Select embedding model: openai:text-embedding-3-small
> Set chunk size: 2000
> Add document paths: tmp/files/paul_graham_essay.txt
✨ Saved rag to '<user-config-dir>/aichat/rags/test5.bin'

#test1> What did the author do growing up?
The author mainly focused on writing and programming growing up ...

.set - adjust settings (non-persistent)

.set max_output_tokens 4096
.set temperature 1.2
.set top_p 0.8
.set rag_top_k 4
.set function_calling true
.set compress_threshold 1000
.set dry_run true

Local Server

AIChat comes with a built-in lightweight web server.

$ aichat --serve
Chat Completions API:
LLM Playground:

$ aichat --serve  # to specify a different server address

OpenAI format API

AIChat offers the ability to function as a proxy server for all LLMs. This allows you to interact with different LLMs using the familiar OpenAI API format, simplifying the process of accessing and utilizing these LLMs.

Test with curl:

curl -X POST -H "Content-Type: application/json" -d '{

LLM Playground

The LLM Playground is a webapp that allows you to interact with any LLM supported by AIChat directly in your browser.


LLM Arena

The LLM Arena is a web-based platform where you can compare different LLMs side-by-side.


Function Calling

Function calling supercharges LLMs by connecting them to external tools and data sources. This unlocks a world of possibilities, enabling LLMs to go beyond their core capabilities and tackle a wider range of tasks.

We have created a new repository to help you make the most of this feature:

Here's a glimpse of what function calling can do for you:




Copyright (c) 2023-2024 aichat-developers.

AIChat is made available under the terms of either the MIT License or the Apache License 2.0, at your option.

See the LICENSE-APACHE and LICENSE-MIT files for license details.