ai-ask

• Talks to and Listen ai agents from your terminal as simple as possible.
• Allows you to enter prompt input directly inline in the terminal, or reading from a text file
• Saves code files created by ai agents into files.

Commands:

• ask <prompt to ask ai agent> Ask an inline prompt and get the output text in screen and audio playback
• ask -f <prompt_file> Ask something using file prompt and get the output text in screen and audio playback
• ask -f <prompt_file> -s Ask something using file prompt, save the output (prompt, input and generated files) and get the output text in screen and playback
• ask voices Get a list of all available voices in all languages
• ask voices <lang> Get a list of all available voices in specific language (en, es, de, ...)

If you want to stop playing the voice, press escape. The text will continue to be generated.

Sample prompts

There are few sample prompt files you can try with:

• ask -f ./sample_prompts/todo-generation.txt -s
• ask -f ./sample_prompts/code-generation/model-entity.txt -s
• ask -f ./sample_prompts/code-generation/fix-function.txt -s
• ask -f ./sample_prompts/code-generation/fix-poc.txt -s

Setup instructions:

Install and run ollama:

docker run -d --restart unless-stopped -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama

Install and run openTTS:

docker run -d --restart unless-stopped -it -p 5500:5500 --name opentts synesthesiam/opentts:en --no-espeak

Install speaker npm module globally:

npm install -g speaker

Make 'speaker' global module importable from 'ask' command:

export NODE_PATH=$(npm root -g)

Build and install agent client:

make

This will install a command called ask in /usr/local/bin/ask

By default services are assumed to run in localhost. If you run the services on a different machine, customize the env vars:

In your .bashrc file:

export ASK_OLLAMA_BASEURL=http://localhost:11434
export ASK_OPENTTS_BASEURL=http://localhost:5500
export ASK_DEFAULT_VOICE=coqui-tts:en_vctk#13
export ASK_LLM_MODEL=llama3

or before the ask as:

ASK_DEFAULT_VOICE=coqui-tts:en_vctk#1 ask tell me nice story

(optional) MTLs

If you have an mtls server that acts as a proxy to your ollama and opentts services you can use the MTLs options (in your .bashrc); example::

export ASK_MTLS=true
export ASK_MTLS_CA=/home/myuser/.ask/certs/ca.crt
export ASK_MTLS_CERT_KEY=/home/myuser/.ask/certs/client.key
export ASK_MTLS_CERT=/home/myuser/.ask/certs/client.crt
export ASK_OLLAMA_BASEURL=http://localhost:11434/ollama
export ASK_OPENTTS_BASEURL=http://localhost:5500/opentts
export ASK_DEFAULT_VOICE=coqui-tts:en_vctk#1
export ASK_LLM_MODEL=llama3