pAIpe: Your LLM shell assistant

pAIpe is a command-line tool designed to streamline your interactions with Large Language Models (LLMs). It provides a flexible and configurable pipeline to process text input through various LLM providers and models, managed through profiles for ease of use and customization.

Features

• Command-Line Interface: Simple and intuitive CLI for interacting with LLMs directly from your terminal.
• Profile-Based Configuration: Define and manage multiple LLM configurations (profiles) in a paipe.yaml file, allowing you to switch between different providers, models, and settings effortlessly.
• Streaming Output: Supports streaming responses from LLMs for a more interactive and real-time experience.
• Input from Stdin or File: Read input text from standard input for quick prompts or from files for processing larger documents.
• Extendable Configurations: Designed to be easily extended with new LLM providers and models by modifying the configuration file.

Installation

# install with openai protocol supported
python -m pip install -U paipe[openai]
# install with all protocol supported
python -m pip install -U paipe[all]

Or install using uv(run with uvx paipe):

# install with openai protocol supported
uv tool install -U paipe[openai]
# install with all protocol supported
uv tool install -U paipe[all]

Configuration

paipe is configured using a paipe.yaml file. This file should be placed in one of the following locations (in order of priority):

1. Current directory (./paipe.yaml)
2. User local directory (~/.local/share/paipe/paipe.yaml)
3. System configuration directory (/etc/paipe.yaml)

If no paipe.yaml is found in these locations, paipe will prompt an error message and exit.

Here's an example of a paipe.yaml configuration file(access api via openrouter):

default:
    protocol: openai
    api_key: 'YOUR_OPENROUTER_API_KEY_HERE' # Replace with your actual API key
    base_url: 'https://openrouter.ai/api/v1'
    model: 'deepseek/deepseek-chat:free'
deepseek-r1:
    protocol: openai
    api_key: 'YOUR_OPENROUTER_API_KEY_HERE' # Replace with your actual API key
    base_url: 'https://openrouter.ai/api/v1'
    model: 'deepseek/deepseek-r1:free'

Each section in the paipe.yaml file defines a profile. Here's a breakdown of the profile settings:

• protocol: Specifies the LLM API protocol. Currently, paipe is configured to use openai which can interface with various providers through platforms like OpenRouter or directly with specific APIs.
• api_key: Your API key for the chosen provider. Replace the placeholder 'YOUR_API_KEY_HERE' with your actual API key.
• base_url: The API base URL for the provider. For OpenRouter, it's typically 'https://openrouter.ai/api/v1'.
• model: The specific LLM model to use from the provider. The available models depend on your chosen provider and API key access. Examples include 'google/gemini-2.0-pro-exp-02-05:free', 'qwen2-72b-instruct', 'deepseek/deepseek-r1:free', etc.

Usage

Basic usage:

paipe <prompt>

This will execute paipe with the <prompt> you provide, using the default profile defined in your paipe.yaml and stream the response to your terminal.

cat essay.txt | paipe --profile deepseek-r1 "write a summary of the following text:"

This will read the content of essay.txt and send it to the deepseek-r1 profile for processing.

paipe --list

This will list all the available profiles defined in your paipe.yaml file.

License

paipe is released under the MIT License. See the LICENSE file for more details.