🤖 Pasquale, Your Local Grammar Checking Bot

Tired of typos sneaking into your writing? This repository contains everything you need to run your own, locally hosted, AI-powered grammar, style, and spelling assistant.

No need to send your text off into the cloud—keep everything private and on your own machine. 🏡

 

Using Pasquale with TeXstudio:

pasquale0.1.mp4

Note: This example required less than 1.5GB of VRAM.

---

 

✨ Features ✨

• Local-first and private: All processing happens on your machine. Using local LLMs, your text stays yours.

• AI-powered corrections: Uses modern language models to catch grammar, spelling, and style issues beyond traditional checkers.

• Explanations included: Suggestions come with reasoning, so you can understand and decide what to accept.

• Lightweight by design: Built to work well with small, efficient LLMs without consuming excessive resources.

• Multi-language support (in progress): Ongoing work to support multiple languages effectively, even with smaller models.

 

📜 Prerequisites 📜

• Python 3.7 or newer
  Check with python --version or python3 --version.

• Local LLM server and model
  Any server compatible with the OpenAI API format will work.

• Python dependencies

  • Flask
  • OpenAI package

 

📝 Running Pasquale 📝

🛠️ Setup (Linux command line)

1. Install an LLM inference server and model

   You’ll need a local LLM backend. Popular options include:

   • ollama
   • llama.cpp
   • LM Studio

   Example for gemma3 1B using ollama:

   # install ollama
   curl -fsSL https://ollama.com/install.sh | sh
   
   # download model
   ollama pull gemma3:1b-it-qat

2. Clone the repository

   git clone https://github.com/joaopaulo7/pasquale.git
   cd pasquale

3. Install dependencies

   pip install -r requirements.txt

 

🙋‍♀️ Using Pasquale

1. Start the server

   Either run directly with python3 command:

   python3 src/simple_server.py

   or use a docker container:

   docker build -t pasquale . && docker run --rm -p 5000:5000 pasquale

   You should see something like: Running on http://127.0.0.1:5000

   Note: The server runs over HTTP (not HTTPS), which is fine for local use.

2. Configure credentials and inference options

   Access http://127.0.0.1:5000/config (or directly edit your config.yaml before running) to change the default options:

   • Credentials:
     • base_url: URL to an openAI-compatible endpoint
     • api_key: API key for your LLM server
   • Inference options:
     • model: Model on which to run the bot
     • genres: Desired writing styles
     • extra_prompt: Additional instructions for the model
     • thinking: enable or disable thinking (disabled for faster results)
     • temperature: Controls randomness
     • max_tokens: Limits response length

3. Connect a client

   Pasquale is API-compatible with LanguageTool, so most editors can use it with minimal setup.

   Just replace the LanguageTool server URL with your local Pasquale URL.

   Example (TeXstudio):

   • Go to Options > Configure TeXstudio > Language Checking
   • Set the server URL to: http://localhost:5000/

   You should start seeing suggestions as you type.

 

🏗️ Work in Progress 🏗️

Pasquale is an early-stage, single-developer project—expect frequent changes.

Planned improvements:

• Evaluate and improve integration with other clients
• Expanded language support
• More model-specific prompting

 

---

God, I hope there aren't any typos in this README.