Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/rainu/ask-mai

A scriptable and powerful UI-Chat-Application for chatting with most Gen-AI LLMs
https://github.com/rainu/ask-mai

anthropic app deepseek desktop-app gemini genai golang llm mcp-client mcp-server mistral ollama openai scriptable ui vue3 wails

Last synced: 4 months ago
JSON representation

A scriptable and powerful UI-Chat-Application for chatting with most Gen-AI LLMs

Awesome Lists containing this project

README 

          
# Ask m' AI



**Ask m' AI** (ask my ai -> ask m' ai) is a little Desktop-Chat-Application for LargeLanguageModels (e.g. OpenAI's GPT).



Unlike many chat applications out there, this application aims to be scriptable. 

Which means you can call it from you terminal and gave all necessary options as arguments. 

The conversations will also be printed out in the terminal, so you can use it in your scripts. 



![](demo.png)



https://github.com/user-attachments/assets/a6d16332-55a1-4866-9f3e-31490a488935



## Quickstart



### LocalAI



```shell

ask-mai -b localai --llm.localai.base-url=http://localhost:8080/v1 --llm.localai.model=gpt-4

```



### OpenAI



```shell

ask-mai -b openai --llm.openai.api-key.plain=YOUR_OPENAI_API_KEY

```



### Ollama



```shell

ask-mai -b ollama --llm.ollama.server-url=http://localhost:11434 --llm.ollama.model=llama3.2

```



### Anthropic



```shell

ask-mai -b anthropic --llm.anthropic.api-key.plain=YOUR_ANTHROPIC_API_KEY

```



### Mistral



```shell

ask-mai -b mistral --llm.mistral.api-key.plain=YOUR_MISTRAL_API_KEY --llm.mistral.model=mistral-small-latest

```



### DeepSeek



```shell

ask-mai -b deepseek --llm.deepseek.api-key.plain=YOUR_DEEPSEEK_API_KEY

```



### Google Gemini



```shell

ask-mai -b google --llm.google.api-key.plain=YOUR_GOOGLE_API_KEY

```



## Features



* Support different LLM provider

  * [Github Copilot](https://github.com/features/copilot)

  * [OpenAI](https://openai.com)

  * [LocalAI](https://localai.io/)

  * [AnythingLLM](https://anythingllm.com/)

  * [Ollama](https://ollama.com/)

  * [Mistral](https://mistral.ai/)

  * [Anthropic](https://www.anthropic.com/)

  * [Google AI](https://ai.google.dev//)

* Tool Support

  * You can define your own tools which can be called from the LLM

  * There are some built-in tools:

    * "getSystemInformation" - Get some information about the system

    * "getEnvironment" - Read environment variables from the system

    * "getSystemTime" - Get the current system time

    * "getStats" - Get stats of a file or directory

    * "changeMode" - Changes the mode of file or directory

    * "changeOwner" - Changes the owner of file or directory

    * "changeTimes" - Changes the access and/or modification time of file or directory

    * "createFile" - Create a file on the system

    * "createTempFile" - Create a temporary file on the system

    * "appendFile" - Append content to an existing file

    * "readTextFile" - Read a text file from the system

    * "deleteFile" - Delete a file from the system

    * "createDirectory" - Creates a new directory on the system

    * "createTempDirectory" - Creates a new temporary directory on the system

    * "deleteDirectory" - Delete a directory from the system

    * "executeCommand" - Execute a command on the system

    * "callHttp" - Do a http call

* Run as [MCP](https://www.anthropic.com/news/model-context-protocol)-Server (see below)

  * Which allows you to use ask-mai' tools (even **your** custom tools) for other applications which support MCP-Server

* [MCP](https://www.anthropic.com/news/model-context-protocol)-Server support

  * You can define multiple MCP-Servers which can be used by the LLM (currently only tools)

* Scriptable

  * All settings can be set via:

    * yaml configuration file 

    * environment variables

    * command line arguments

  * The users questions and models answers will be printed out in the terminal

* Customizable

  * Choose between two themes (light and dark)

  * Choose between different languages (english, german)

  * Choose between different code themes (see help for all available themes)



## Install this application



### AUR (Arch Linux)



```bash

yay -S ask-mai

```



## Download this application



You can download the latest version of this application from the [releases page](https://github.com/rainu/ask-mai/releases).

There are many different versions for different operating systems, architectures and feature sets.



The file names are structured as follows:

```

ask-mai-${OS}-${ARCH}-${FEATURE}.${EXTENSION}

```



Available OS:

* darwin - MacOS

* linux - Linux

* windows - Windows



Available Architectures:

* amd64 - 64bit

* 386 - 32bit

* arm64 - ARM 64bit



Available Features:

* compressed - The binary is compressed (can be problematic for some antivirus software - especially on windows)

* console - The binary is console application (only for windows)

* debug - The binary contains devtools (you can inspect the GUI-Sources)



## Start this application as mcp-server



You can start this application as an MCP server by using the `ask-mai-mcp-server` as first positional argument:

```bash

ask-mai ask-mai-mcp-server

```



Or you can rename the binary to `ask-mai-mcp-server` and start it like this:

```bash

ask-mai-mcp-server

```



## How to build this application



1. Install dependencies [see wails documentation](https://wails.io/docs/gettingstarted/installation)

2. Build the application:

```sh

wails build

```



## How to build and install this application (flatpak) by yourself



```

flatpak-builder .flatpak-build de.rainu.ask-mai.yml --repo=.flatpak-repo --install-deps-from=flathub --force-clean --default-branch=master --arch=x86_64 --ccache

flatpak build-bundle .flatpak-repo ask-mai.flatpak --runtime-repo=https://flathub.org/repo/flathub.flatpakrepo --arch=x86_64 de.rainu.ask-mai master

sudo flatpak install --reinstall ask-mai.flatpak

```



## Application starts not fast enough



If the application does not start fast enough for your needs, you can get the prompt by your own and use the `-ui-prompt` flag to start the application with that prompt. 

Here is an example with [rofi](https://github.com/davatorium/rofi):



```sh

#!/bin/sh



PROMPT=$(rofi -dmenu -p "Ask-mAI" -l 0 -location 2)

if [ $? -ne 0 ]; then

    exit 1

fi



ask-mai -ui-prompt "${PROMPT}"

```



## Contributing



If you use [nix](https://nixos.org/), you can use `nix develop` to enter a shell with all dependencies you need for contributing. The flake also utilises [treefmt](https://github.com/numtide/treefmt-nix) to format all *.nix and *.go files. This can be done with `nix fmt` and checked with `nix flake check`.