chatcortex 0.1.0

A privacy-first, local document chatbot engine

ChatCortex

ChatCortex is a privacy-first, local document chatbot engine designed to be embedded into any application.
It enables teams to build document-aware conversational systems without relying on cloud services, while keeping all data within their own infrastructure.

ChatCortex is library-first, API-agnostic, and designed for production and enterprise use cases.

---

Overview

Most document-based chat systems are tightly coupled to user interfaces, depend on external cloud services, or are difficult to integrate into existing platforms.

ChatCortex focuses exclusively on the core infrastructure layer, allowing developers to integrate document chat capabilities into their own systems with minimal overhead.

---

Core Principles

• Privacy-first: All ingestion, retrieval, and inference occur locally
• Library-first design: Use directly as a Python SDK or wrap with custom APIs
• Frontend-agnostic: Compatible with any UI or client application
• Backend-oriented: Built as infrastructure, not a demo application
• Simple and stable API: Small public surface area intended for long-term stability

---

Use Cases

ChatCortex can be used to build:

• Internal enterprise document chatbots
• Knowledge base assistants
• Policy, HR, and legal document Q&A systems
• On-premise or air-gapped AI assistants
• Custom chat APIs for proprietary frontends

---

Quick Start

pip install chatcortex

from chatcortex import ChatCortexEngine

engine = ChatCortexEngine()

# Ingest documents (v0.1 supports .txt files)
engine.ingest("./docs")

# Ask a question
response = engine.chat("What is covered in these documents?")
print(response)

---

Public API (v0.1)

ChatCortexEngine(
    config: ChatCortexConfig | None = None
)

engine.ingest(source: str)
engine.chat(message: str, session_id: Optional[str] = None) -> str
engine.reset_session(session_id: str)

---

Configuration

from chatcortex.config import ChatCortexConfig
from chatcortex import ChatCortexEngine

config = ChatCortexConfig(
    chunk_size=500,
    chunk_overlap=100
)

engine = ChatCortexEngine(config)

---

Design Philosophy

ChatCortex enforces a clear separation of responsibilities:

Layer	Responsibility
Core Engine	Document ingestion, retrieval, session management
API Layer	Optional adapters (REST, gRPC, etc.)
Frontend	Fully owned by the integrator

You are free to:

Use ChatCortex directly as a Python library

Wrap it with your own API layer

Integrate it into existing systems

Ignore any optional adapters

---

Roadmap

v0.1 -> Core engine -> Local document ingestion -> Session handling

Planned

PDF ingestion -> Embeddings and vector database support (FAISS, Chroma) -> Streaming responses -> External connectors (S3, SharePoint) -> Optional FastAPI service wrapper

---

Development Setup

git clone https://github.com/yourusername/chatcortex
cd chatcortex
python -m venv .venv
source .venv/bin/activate
pip install -e .

---

License

MIT License

You are free to use, modify, and integrate ChatCortex into commercial and internal products.

---

Project Goals

ChatCortex is built as infrastructure, not a demo application.

It is intended for teams that require a reusable, local-first document chatbot engine without repeatedly rebuilding the same backend components.