Bibliophile Assistant

A CLI tool that uses Ollama and ChromaDB to create a document-based AI assistant. Point it at folders with documents (Word, PDF, Markdown, Excel, PowerPoint) and ask questions in plain English.

Features

• Document Processing: Supports PDF, Word (DOCX, DOC), Markdown, Text, Excel (XLSX, XLS), and PowerPoint (PPTX, PPT) files
• Ollama Integration: Uses local Ollama models for embeddings and chat
• ChromaDB: Vector database for efficient document retrieval
• Auto-Setup: Detects hardware (RAM, GPU) and suggests appropriate models
• User-Friendly CLI: Easy-to-use command-line interface with progress indicators

---

Quick Start (No Python Required)

1. Install

Windows (PowerShell) — one-liner, works immediately:

$env:BIBLIOPHILE_VERSION='v0.1.3'; irm https://raw.githubusercontent.com/JyslaFancy/Bibliophile-Assistant/main/install.ps1 | iex

This downloads the latest bibliophile.exe, puts it in %LOCALAPPDATA%\Programs\Bibliophile, and adds it to your PATH.

After it finishes, restart your terminal and run:

bibliophile setup --install-ollama --pull-models --start-server

Windows (winget) — once approved by winget-pkgs:

winget install JyslaFancy.BibliophileAssistant

Currently pending submission. Use the PowerShell install above until then.

Linux — one-liner:

curl -fsSL https://raw.githubusercontent.com/JyslaFancy/Bibliophile-Assistant/main/install.sh | bash

Or download the binary directly from the Releases page.

2. Run

# Windows (PowerShell / Command Prompt)
.\bibliophile.exe setup --install-ollama --pull-models --start-server

# Linux
./bibliophile setup --install-ollama --pull-models --start-server

That's it. No Python, no pip, no PATH — just download and run.

3. Index and Ask

# Index your documents
./bibliophile index /path/to/your/documents

# Ask a question
./bibliophile ask "What is this about?"

# Start interactive chat
./bibliophile chat

Prerequisites: You'll need Ollama installed (the setup command above handles this automatically). The standalone binary bundles everything else — Python, all libraries, ChromaDB — so you don't install anything else.

---

Usage

Setup

# Auto-detect hardware and suggest models
bibliophile setup

# Full auto-setup (install Ollama + pull models + start server)
bibliophile setup --install-ollama --pull-models --start-server

Index Documents

# Index all documents in a folder
bibliophile index /path/to/documents

# With a custom collection name
bibliophile index /path/to/documents --name my_project

# With custom chunk size
bibliophile index /path/to/documents --chunk-size 500

# Overwrite existing collection
bibliophile index /path/to/documents --overwrite

Ask Questions

bibliophile ask "What is the main topic of these documents?"
bibliophile ask "What are the key findings?" --collection my_project
bibliophile ask "Summarize the documents" --limit 10

Interactive Chat

bibliophile chat
bibliophile chat --collection my_project

Chat Commands:

• /quit, /exit, /q - End the chat session
• /help, /h - Show available commands
• /clear, /reset - Clear conversation history
• /collection, /col - Show current collection

Manage Collections

bibliophile list-collections
bibliophile delete-collection my_project

Configuration

bibliophile config-show         # Show current configuration
bibliophile config-reset        # Reset to defaults

---

Supported File Types

• .pdf — PDF documents
• .docx — Word documents (Office Open XML)
• .doc — Word documents (legacy, requires antiword or catdoc)
• .md — Markdown files
• .txt — Plain text files
• .xlsx / .xls — Excel spreadsheets
• .pptx / .ppt — PowerPoint presentations

---

Hardware Requirements

Bibliophile automatically detects your hardware and suggests appropriate models:

• Minimum: 8GB RAM — uses lightweight models
• Recommended: 16GB+ RAM — comfortable performance
• With GPU: 12GB+ VRAM — GPU-accelerated inference

---

For Developers

Install from PyPI

pip install bibliophile-assistant

Install from Source

git clone https://github.com/JyslaFancy/Bibliophile-Assistant.git
cd Bibliophile-Assistant
pip install -e ".[dev]"

Build the Standalone Executable

pip install -e ".[dev]"
python scripts/build.py

Output goes to dist/bibliophile (or dist/bibliophile.exe on Windows).

Submitting to winget-pkgs

After creating a release, generate the winget manifest:

python scripts/generate_winget.py v0.1.1

Then submit the generated winget/ files to the winget-pkgs community repository as a pull request. Once merged, users can run:

winget install JyslaFancy.BibliophileAssistant

Configuration

Configuration is stored in ~/.bibliophile/config.yaml:

ollama:
  chat_model: "llama3"
  embedding_model: "nomic-embed-text"
  base_url: "http://localhost:11434"

chroma:
  path: ".bibliophile/chroma"

document_processing:
  chunk_size: 1000
  chunk_overlap: 200
  supported_extensions:
    - ".pdf"
    - ".docx"
    - ".doc"
    - ".md"
    - ".txt"
    - ".xlsx"
    - ".xls"
    - ".pptx"
    - ".ppt"

License

MIT License — See LICENSE for details.