Nolara is a locally-run AI assistant

Powered by Ollama and cutting-edge open models. Designed for privacy, speed, and versatility, Nolara offers intelligent, offline-friendly assistance without relying on the cloud. Whether you’re coding, creating, or just curious, Nolara is your ever-present, whisper-smart companion—right on your machine.

1. Prerequisite

1.1. Install ollama

https://ollama.com

1.2 Build nolara app

cd Nolara
./build.bash

1.3 Run nolara app

nolara

Other examples:

nolara --help
cat README | nolara -p "Summarize please"

Then you have system wide nolara command available.

Customzation

💬 system_prompts

Add custom system prompts under

(.venv) ➜  Nolara git:(main) ✗ ll ~/.nolara/system_prompts/
total 16
-rw-r--r--@ 1 usr  staff    52B May 31 16:39 basic.txt
-rw-r--r--@ 1 usr  staff    68B May 31 12:52 short.txt

🛠️ tools

Add custom tools under

(.venv) ➜  Nolara git:(main) ✗ ll ~/.nolara/tools/
total 8
-rw-r--r--@ 1 usr  staff   606B May 31 18:14 basic.py
-rw-r--r--@ 1 usr  staff     0B May 31 18:34 websearch.py

Configuration

cat ~/.nolara/config.json

{
  "tools": ["*", "!micros_tools.py"],
  "language":"en-US",
  "models": ["qwen3:4b", "llama3.2:1b", "gemma2:latest"]
}

tools: Select tools for agentic models (Nolara/tools/*.py), * or all means select all tools. Disable tool with !. You can add modules individually as well.

language: Under development

models: required models with autoinstall (ollama pull)

Nolara as a command line tool (beta)

(.venv) ➜  Nolara git:(main) ✗ cat README.md | ./main.py -p "Summarize the content in few sentences"

.../usr_config.json already exists
Check config updates...
	No updates needed.
[1/2] Model qwen3:4b already available.
[2/2] Model gemma2:latest already available.

Processing...

Nolara is a privacy-focused, open-source AI assistant powered by Ollama and
cutting-edge models. It runs locally on your machine without relying on the
cloud, offering intelligent assistance for coding, creative tasks, and general
inquiries.

To use Nolara, you need to install Ollama, set up a Python virtual environment,
install required dependencies, and download the desired models. You can then
start the Nolara chat interface and customize it with system prompts and tools.
The project also provides configuration options for selecting specific models
and tools.

Model handling TL;DR

Manual install:

Small profile example

ollama pull gemma2:latest
ollama pull gemma3:4b
ollama pull qwen3:4b

Medium profile example

ollama pull gemma2:latest
ollama pull gemma3:12b
ollama pull qwen3:8b

Model info

Current promising models

gemma2 details

gemma3 details

qwen3 details

ollama list

gemma2:latest              ff02c3702f32    5.4 GB
gemma3:12b                 f4031aab637d    8.1 GB

Reasoning models + Tools

qwen3:14b                  7d7da67570e2    9.3 GB

Development envoronment

Setting Up a Python Virtual Environment and Installing Requirements

Follow these steps to create a virtual environment using virtualenv, install dependencies from requirements.txt, and activate the environment.

cd Nolara/

# Create
python3 -m venv .venv

# Activate
source .venv/bin/activate

# Install base dependencies
pip install -r ./nolara/requirements/requirements-base.txt

# Install Audio dependencies (optional)
pip install -r ./nolara/requirements/requirements-audio.txt

This command you will need to activate the virtual environment: source .venv/bin/activate

Start the Nolara chat interface

cd Nolara
source .venv/bin/activate
./nolara/main.py

---

---

Additional models for Ollama Coding Buddy

Coding models (Continue plugin)

continue.dev

codellama:13b              9f438cb9cd58    7.4 GB
qwen2.5-coder:7b           2b0496514337    4.7 GB

Embedding:
nomic-embed-text:latest    0a109f422b47    274 MB

git push -u origin main