kernelmind 0.1.1

A minimalist thinking kernel for LLM agents

KernelMind

The minimalist thinking kernel for building LLM-powered AI agents.

KernelMind is a lightweight, flexible framework designed to build agentic LLM systems using a simple yet powerful abstraction: Point + Line.
It is ideal for rapidly prototyping workflows, agents, RAG pipelines, and structured thinking systems — all in under 100 lines of core code.

---

✨ Key Features

• 🧠 Agentic-first architecture – Built from the ground up for multi-step reasoning
• 🔁 Point + Line abstraction – Minimal and composable flow control
• ⚙️ Retry / Fallback – Built-in error recovery mechanism
• 🚀 Sync & Async support – Automatically detects coroutine behavior
• 🧪 Testable by design – Small units, observable memory store
• 🌿 Extensible – Implement any agent/workflow pattern

---

💡 Core Concepts

Concept	Description
Point	A minimal unit of logic (LLM or utility call)
Line	A composition of points connected via actions
Memory	A dictionary-style global store shared across points
load() / process() / save()	The 3 lifecycle steps of each Point

class Point:
    def load(self, memory): ...
    def process(self, item): ...
    def save(self, memory, input, output): ...

---

📦 Installation

Option 1: Using uv (Recommended)

uv is a fast Python package installer and resolver, written in Rust.

1. Install uv

# Using pipx (recommended)
pipx install uv

# Or using Homebrew (macOS)
brew install uv

# Or using curl
curl -LsSf https://astral.sh/uv/install.sh | sh

2. Create virtual environment and install dependencies

# Clone the repository
git clone https://github.com/your-org/kernelmind-framework.git
cd kernelmind-framework

# Create virtual environment and install in development mode
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

# Or install specific dependencies
uv add requests
uv add --dev pytest

3. Run the example

python examples/agent_qa.py

Option 2: Using pip (Traditional)

# Install from PyPI (when published)
pip install kernelmind

# Or install in development mode
git clone https://github.com/your-org/kernelmind-framework.git
cd kernelmind-framework
pip install -e .

Option 3: Manual Setup

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt
pip install -e .

---

🚀 Quick Example: Q&A Agent

from kernelmind.core import Point, Line

class GetQuestion(Point):
    def load(self, memory):
        return "start"  # Return a non-None value to trigger process()
    
    def process(self, _): 
        user_input = input("Ask: ")
        return user_input
    
    def save(self, memory, _, out): 
        memory["question"] = out; 
        return "default"

class AnswerQuestion(Point):
    def load(self, memory): 
        return memory["question"]
    
    def process(self, q): 
        return f"Answer: {q}"
    
    def save(self, memory, _, out): 
        print("Answer:", out)

ask = GetQuestion()
answer = AnswerQuestion()
ask >> answer
qa_line = Line(entry=ask)
qa_line.run({})

---

🧠 Supported Design Patterns

• ✅ Multi-step Workflow
• ✅ Agent with Contextual Actions
• ✅ RAG (Retrieval-Augmented Generation)
• ✅ Map Reduce
• ✅ Structured YAML Output

See docs/guide.md for examples and best practices.

---

📁 Project Structure

kernelmind/
├── core.py          # Core logic: Point + Line
├── utils/           # External API wrappers (LLM, search, etc.)
├── examples/        # Use cases (Agent, RAG, Workflow)
├── tests/           # Unit tests
├── docs/guide.md    # Developer documentation

---

📖 Documentation

Check out the full usage guide and design patterns in:

📚 docs/guide.md

---

🧪 Run Tests

# Using uv
uv run pytest tests/

# Using pip
pytest tests/

---

❤️ Philosophy

From Points to Minds. From Lines to Agents.

KernelMind is not just a framework.
It is a philosophy of thinking through structure — building powerful systems with minimal building blocks.

---

📬 Contribution

Pull requests welcome! If you have suggestions, open an issue or start a discussion.

MIT License · Built for the AI-native future.