pai-repl 0.1.13

Python REPL with LLM integration

pai: An AI agent inside your Python REPL

A Python REPL with a built in AI agent and code generation.

• REPL history is used for LLM context.
• All code can be edited or cancelled before execution
• Runs locally on your machine, so it has full system and internet access
• Supports OpenAI and llama.cpp

Installation

pip install pai-repl

Usage

When you invoke pai, it will start an interactive Python REPL.

$ pai
INP [0]> nums = [1,2,3]
INP [1]> nums[0]
OUT [1]> 1
INP [2]>

Generate code with gen: <prompt>. The generated code will be displayed and you can accept, edit or cancel it.

INP [0]> nums = [1,2,3]
INP [1]> gen: average nums
LLM [1]>
# to find the average of the numbers, we sum all the elements and then divide by the number of elements

average_nums = sum(nums) / len(nums)
average_nums

OK? [1]> # to find the average of the numbers, we sum all the elements and then
    ...> divide by the number of elements
    ...>
    ...> average_nums = sum(nums) / len(nums)
    ...> average_nums
OUT [1]> 2.0
INP [2]>

Start the agent with pai: <prompt>. The agent will continuously generate and run code until it completes the task or fails.

INP [0]> pai: pick 2 random wikipedia articles and tell me how they are related
LLM [0]>
# Let's use the wikipedia API to fetch random articles
import requests
import json

def get_random_wikipedia_articles(count):
    S = requests.Session()
    ...

Prompt pai from the command line

$ pai "find the largest file in the current directory"

Configuration

The default model is OpenAI GPT-4. You will need to set your OpenAI API key.

$ export OPENAI_API_KEY=<your key>
$ pai
pai v0.1.12 using gpt-4
'Ctrl+D' to exit. 'Ctrl+o' to insert a newline.
INP [0]>

Specify OpenAI model

$ pai --openai gpt-3.5-turbo

Alternatively, you can use llama.cpp compatible models

$ pai --llama <path to model>