Skip to content

Instantly share code, notes, and snippets.

View vitorcalvi's full-sized avatar
🎯
Focusing

Carlos Vitor Botti Calvi vitorcalvi

🎯
Focusing
View GitHub Profile

Ranked Model Tiers - Free Models from nano-gpt.com

S-Tier (Elite - Highest Capability)

Rank Model Provider Strength
1 deepseek-v3.2:thinking DeepSeek Best reasoning, cost-effective
2 claude-opus-4:thinking Anthropic Top-tier reasoning & analysis
3 deepseek-v3.2 DeepSeek Overall best value/performance
4 gemini-2.5-pro:thinking Google Strong reasoning, multimodal
@vitorcalvi
vitorcalvi / README.md
Created January 15, 2026 15:09
Grok + Claude 4.5 Supervisor System - Dual-agent AI system with real-time bug detection and correction

Grok + Claude 4.5 Supervisor System

Overview

A dual-agent AI system where Grok writes code quickly while Claude 4.5 supervises, tests, and corrects bugs in real-time.

Architecture

┌─────────────────┐         ┌──────────────────┐
@vitorcalvi
vitorcalvi / acp-inter-agent-communication.md
Last active January 15, 2026 13:15
Complete guide for inter-agent communication between Charm Crush and OpenCode using the Agent Communication Protocol (ACP)

Full Inter-Agent Communication: Charm Crush ↔ OpenCode via ACP

There are two different ACP protocols - be careful not to confuse them:

Protocol Purpose Transport
OpenCode ACP (Agent Client Protocol) Editor-to-agent integration JSON-RPC via stdio
ACP (Agent Communication Protocol) Agent-to-agent interoperability REST/HTTP

For Charm Crush ↔ OpenCode communication, use the Agent Communication Protocol (ACP) developed by IBM/BeeAI under Linux Foundation.

@vitorcalvi
vitorcalvi / alpine-setup.sh
Last active December 30, 2025 15:19
Alpine Linux AMD GPU Workstation Setup Script
#!/bin/sh
# Alpine Linux System Setup Script - V2 (ROBUST FIX)
# Target: AMD Ryzen 7 H 255 (Zen 4) / Radeon 780M
# Alpine Linux v3.23+
set -e
echo "=== Alpine Linux System Setup ==="
echo "This script will install and configure packages for an AMD Ryzen APU workstation"
echo ""
@vitorcalvi
vitorcalvi / docker-compose.yml
Created December 30, 2025 07:00
Ollama Rocm Dokploy
version: "3.8"
services:
ollama:
image: ollama/ollama:rocm
restart: unless-stopped
privileged: true
dns:
- 8.8.8.8
- 1.1.1.1
@vitorcalvi
vitorcalvi / CyberDog1_Disassembly_Image_Prompts.md
Created December 20, 2025 08:26
Xiaomi CyberDog 1 Disassembly Prompts

REFINED LLM Prompts: Xiaomi CyberDog 1 Disassembly Manual

Optimized for Gemini Imagen 3 - Tested & Validated


Image 1: Exterior Overview - Port Locations ✅ TESTED

Prompt:

Technical diagram of Xiaomi CyberDog 1 quadruped robot, side perspective view, clean white background. The robot body is matte black with smooth surfaces and four legs. Red arrows with labels pointing to key locations: "Power Button" on top back panel, "Touch Sensor" on center back panel, "Download USB-C Port" on left side near front leg, "Charging Port" on right side near rear leg. Isometric 3D perspective view. Technical instruction manual style. High contrast red arrows against black robot body. Sans-serif typography for labels.
@vitorcalvi
vitorcalvi / Build Pay-to-Reveal SaaS.md
Created December 11, 2025 10:33
Build Pay-to-Reveal SaaS

PROMPT: Build Pay-to-Reveal SaaS MVP (Self-Hosted + AI Moderation)

0. Non-Negotiable Constraints (Read First)

You are building a real, production-ready MVP that I will host on my own server.

  • Hosting / runtime
    • Self-hosted on my Linux home server (Node.js).
    • Must run with:
  • npm install
Let’s say: “Build a full FastAPI backend + React frontend for a new SaaS.”
Design (ChatGPT Pro)
You: “Design architecture: entities, routes, DB schema, auth flow, deployment strategy.”
ChatGPT gives you:
Entity list
@vitorcalvi
vitorcalvi / install_Kokoro-FastAPI.sh
Created December 11, 2025 08:54
Kokoro-FastAPI Voice generator
#!/usr/bin/env bash
# Simple one-shot runner for Kokoro-FastAPI on macOS
set -e
# 1) Go to project folder
cd "${HOME}/Desktop/Kokoro-FastAPI"
# 2) (Optional) Create and use a virtualenv
if [ ! -d ".venv" ]; then
@vitorcalvi
vitorcalvi / Universal Adaptive Strategy V4 (ATR-Based).pine
Last active December 11, 2025 08:46
from 15s to 30m timeframe all is positive PnL
//@version=5
strategy("Universal Adaptive Strategy V4 (ATR-Based)", overlay=true, process_orders_on_close=true)
// --- 1. SETTINGS & INPUTS ---
// TREND FILTER (The "Big Boss")
// We only trade if price is above this EMA. Keeps you out of crashes.
ema_len = input.int(200, "Trend Filter (EMA)", group="Trend")
// SUPERTREND (Entry Signal)