npx agentspec-cli install 1cd9ff27-31c9-4f6d-9c80-1b3e1445830d --framework ClaudeCode→ .claude/skills/📋 Project Overview/SKILL.md
## 📋 Project Overview
SentidoFinanciero is a complete full-stack application that uses artificial intelligence to automatically categorize and analyze bank statement transactions. Upload PDF statements and get intelligent insights about your spending patterns using local AI processing.
## 🏗 Architecture
**Stack**: React + FastAPI + PostgreSQL + Ollama (Local LLM)
```
Frontend (React) ←→ Backend (FastAPI) ←→ Database (PostgreSQL)
↓
Ollama LLM Service
(Llama 3.2 1B Model)
```
## 🤖 AI Categorization System
**3-Tier Hybrid Classification**:
1. **Tier 1: Exact Keyword Matching** (80% of transactions, <1ms)
```python
"OXXO ROMA" → "alimentacion" (Confidence: 1.0)
"PEMEX 1234" → "gasolineras" (Confidence: 1.0)
```
2. **Tier 2: Pattern Recognition** (15% of transactions, ~1ms)
```python
"REST BRAVA" → regex: r'\brest\b' → "alimentacion" (Confidence: 0.8)
"DR MARTINEZ" → regex: r'\bdr\s+[a-z]+' → "salud" (Confidence: 0.8)
```
3. **Tier 3: LLM Analysis** (5% of transactions, 100-500ms)
```python
"POINTMP*VONDYMEXICO" → Llama 3.2 1B → "servicios" (Confidence: 0.7)
```
**Supported Categories**: Alimentación, Gasolineras, Servicios, Salud, Transporte, Entretenimiento, Ropa, Educación, Transferencias, Seguros, Intereses/Comisiones, Otros
## 🔌 API Endpoints
### Core Endpoints
- `POST /api/v1/statements/upload` - Upload PDF statement
- `POST /api/v1/statements/{id}/process` - Process with AI
- `GET /api/v1/statements` - List all statements
- `GET /api/v1/statements/{id}` - Get statement details
- `GET /api/v1/statements/{id}/transactions` - Get transactions
- `GET /api/v1/statements/{id}/analysis` - Get spending analysis
- `PUT /api/v1/transactions/{id}` - Update transaction category
- `DELETE /api/v1/statements/{id}` - Delete statement
## 🛠 Tech Stack Details
### Frontend
- **React 18** with hooks and modern patterns
- **Vite** for fast development and builds
- **Tailwind CSS** for utility-first styling
- **React Router** for client-side routing
- **React Query** for server state management
- **Chart.js** for interactive data visualizations
- **React Dropzone** for file upload
- **Axios** for HTTP requests
- **Lucide React** for beautiful icons
### Backend
- **FastAPI** with async/await support
- **SQLAlchemy** for database ORM
- **AlembBuilt on the Agent Skills standard
Project Overview
This is **Finance Guru™** - a private AI-powered family office system built on BMAD-CORE™ v6 architecture. This repository serves as the operational center for a multi-agent financial intelligence system that provides research, quantitative analysis, strategic planning, and compliance oversight.
by AgentHub Crawler
Monity Web Application - Cursor Rules
Monity is a full-stack personal finance management application with AI-powered transaction categorization, collaborative expense splitting, and financial health insights. Built with React and Node.js following MVC architecture.
by AgentHub Crawler
.cursorrules for AI and Financial Computing Project
1. Prioritize code readability and maintainability. 2. Ensure all code adheres to PEP 8 (Python) or relevant style guides for other languages. 3. Use meaningful variable and function names that reflect their purpose.
by AgentHub Crawler
Sign in to leave a comment.
Reglas y Contexto del Agente
> Soy Leonard Cuenca Roa, un Ingeniero en Informática con más de 10 años de experiencia en el desarrollo de aplicaciones web full stack (PHP, Python, .NET, Java) y en la gestión de bases de datos SQL y NoSQL. Actualmente, estoy cursando una Maestría en Análisis y Visualización de Datos Masivos en la UNIR - Universidad Rioja de España, para complementar mi experiencia técnica. Mi perfil me permite actuar como un KAM tecnológico, diseñando e implementando soluciones que procesan grandes volúmenes
by AgentHub Crawler