🤖 AI Codebase Intelligence Platform

An AI-powered Codebase Intelligence Platform built with a modular, SOLID-based, multi-provider and multi-model architecture.

The platform is designed to understand software repositories, analyze source code, extract codebase knowledge, and provide AI-powered intelligence over the complete codebase.

Project Status

Current Phase: Phase 0 — General Foundation

Architecture

The platform follows:

• SOLID principles
• Separation of concerns
• Dependency inversion
• Interface-based design
• Modular architecture
• Multi-provider support
• Multi-model support
• Dependency injection
• Service-oriented design
• Extensible AI architecture

AI Providers

The platform is designed to support multiple AI providers through a common provider abstraction.

Configured providers:

1. Ollama
2. OpenAI
3. Anthropic
4. Google Gemini
5. Mistral
6. Groq
7. Cohere
8. DeepSeek

AI Models

The platform uses a centralized provider/model configuration.

Ollama

• qwen2.5:1.5b
• gemma2:2b
• gemma3:4b
• mistral:latest
• phi3:latest
• qwen3:latest
• llama3.1:latest
• llama3:8b
• deepseek-coder:latest

OpenAI

• gpt-5-mini
• gpt-4o
• gpt-4o-mini

Anthropic

• claude-sonnet-4-5
• claude-haiku-4-5
• claude-opus-4-1

Google Gemini

• gemini-3.6-flash
• gemini-2.5-flash
• gemini-2.5-pro

Mistral

• mistral-medium-latest
• mistral-large-latest
• mistral-small-latest

Groq

• llama-3.3-70b-versatile
• llama-3.1-8b-instant
• mixtral-8x7b-32768

Cohere

• command-a-03-2025
• command-r-plus
• command-r

DeepSeek

• deepseek-chat
• deepseek-reasoner
• deepseek-v4-flash

Project Phases

Phase 0 → General Foundation
Phase 1 → Codebase Knowledge Ingestion Process
Phase 2 → Codebase Intelligence & Memory Process
Phase 3 → AI Codebase Interaction Process
Phase 4 → Validation & Advanced Codebase Intelligence

Phase 0 Components

Configuration
Provider Abstraction
Provider Clients
Provider Registry
Model Registry
Provider Factory
Dependency Injection
Application Bootstrap
Streamlit Foundation
Logging
Exception Handling
Testing Foundation

Project Structure

app/
├── application/
├── config/
├── core/
├── domain/
├── infrastructure/
└── presentation/

tests/
├── unit/
└── integration/

docs/
├── architecture.md
└── phase-0.md

Configuration

Create a local .env file from .env.example.

Never commit .env or API keys to source control.

Provider and model configuration is centralized so that AI provider selection remains independent from the core codebase intelligence logic.

Running the Application

Install dependencies:

pip install -r requirements.txt

Start Streamlit:

streamlit run app.py

Development Principles

Every phase follows:

PLAN
 ↓
DESIGN
 ↓
IMPLEMENT
 ↓
TEST
 ↓
FIX
 ↓
STABILIZE
 ↓
DOCUMENT
 ↓
NEXT PHASE

Core Platform Vision

The Codebase Intelligence Platform will progressively provide:

Codebase
   ↓
Repository Discovery
   ↓
Codebase Ingestion
   ↓
Code Understanding
   ↓
Knowledge Extraction
   ↓
AI Intelligence
   ↓
Codebase Interaction
   ↓
Advanced Codebase Intelligence

The goal is to build an AI system capable of understanding a software codebase as a complete system rather than treating individual source files as isolated documents.

License

MIT License

### `docs/architecture.md`

```markdown
# Architecture

## Architectural Layers

```text
Presentation
     ↓
Application
     ↓
Domain
     ↓
Infrastructure

Dependency Direction

Presentation → Application → Domain
                         ↓
                  Infrastructure

The domain layer depends on abstractions rather than concrete infrastructure implementations.

Application Architecture

app.py
  ↓
Application
  ↓
Dependency Injection Container
  ↓
Services / Agents
  ↓
Provider Factory
  ↓
AI Provider
  ↓
AI Model

The Streamlit entry point remains thin and does not directly communicate with individual AI providers.

Provider Architecture

AIProvider
    ↑
Concrete Provider
    ↓
Provider Client
    ↓
External AI API / Local Model

Provider Factory

ProviderFactory
       │
       ├── Ollama
       ├── OpenAI
       ├── Anthropic
       ├── Google Gemini
       ├── Mistral
       ├── Groq
       ├── Cohere
       └── DeepSeek

The Provider Factory is responsible for creating the appropriate provider implementation.

SOLID

Single Responsibility

Components have focused responsibilities.

Examples:

Configuration
    → Application configuration

Provider
    → AI communication

Service
    → Business logic

Agent
    → AI workflow

Presentation
    → UI rendering

Container
    → Dependency management

Open/Closed

New providers, models, services, and intelligence capabilities can be introduced without modifying unrelated core logic.

Liskov Substitution

Concrete providers implement the common provider abstraction and can be substituted wherever the base abstraction is required.

Interface Segregation

Interfaces remain focused and minimal.

Dependency Inversion

Higher-level application and domain components depend on abstractions rather than concrete infrastructure implementations.

Multi-Provider Architecture

8 AI providers are supported through a common provider abstraction.

Multi-Model Architecture

AI models are centrally configured and associated with their respective providers.

Provider and model selection remains independent from the core codebase intelligence business logic.

Codebase Intelligence Architecture

The long-term architecture follows:

Repository
    ↓
Repository Discovery
    ↓
File Discovery
    ↓
Code Parsing
    ↓
Code Metadata
    ↓
Knowledge Extraction
    ↓
Knowledge Store
    ↓
Retrieval
    ↓
AI Intelligence
    ↓
Codebase Interaction

### `docs/phase-0.md`

```markdown
# Phase 0 — General Foundation

## Objective

Establish a clean, modular, SOLID-based foundation for the
AI Codebase Intelligence Platform.

The purpose of Phase 0 is to create the architectural foundation
required for future codebase ingestion, intelligence, memory,
interaction, and advanced analysis capabilities.

## Completed

- Project definition
- Architecture design
- Repository structure
- Configuration foundation
- Provider abstraction
- Provider clients
- Provider registry
- Model registry
- Provider factory
- Dependency injection
- Application bootstrap
- Streamlit foundation
- Logging
- Exception handling
- Testing foundation
- Configuration validation
- Architecture stabilization
- Documentation

## Phase 0 Architecture

```text
Streamlit
    ↓
Presentation
    ↓
Application
    ↓
Dependency Injection Container
    ↓
Services / Agents
    ↓
Provider Factory
    ↓
AI Provider
    ↓
AI Model

Phase 0 Goals

✅ Establish project foundation
✅ Establish clean architecture
✅ Establish SOLID principles
✅ Establish provider abstraction
✅ Establish multi-provider foundation
✅ Establish multi-model foundation
✅ Establish dependency injection
✅ Establish application bootstrap
✅ Establish Streamlit foundation
✅ Establish testing foundation

Status

Phase 0 is completed and ready for Phase 1.

Next Phase

Phase 1 — Codebase Knowledge Ingestion Process

## Phase 0 Tracker

```text
✅ 0.1  Project Definition & Architecture
✅ 0.2  Repository Structure
✅ 0.3  Configuration Foundation
✅ 0.4  Multi-Provider / Multi-Model Foundation
✅ 0.5  Provider Abstraction
✅ 0.6  Provider Clients
✅ 0.7  Provider Registry
✅ 0.8  Model Registry
✅ 0.9  Provider Factory
✅ 0.10 Dependency Injection Container
✅ 0.11 Logging + Exception Foundation
✅ 0.12 Configuration Validation
✅ 0.13 Application Bootstrap
✅ 0.14 Streamlit Foundation
✅ 0.15 Testing Foundation
✅ 0.16 Stabilization
✅ 0.17 Documentation
✅ 0.18 Phase 0 Review
✅ 0.19 Phase 0 Complete

➡️ Next → Phase 1: Codebase Knowledge Ingestion Process