Reverse MCP Servers: Connecting Local Tools to Cloud-Based AI Agents

Learn how to implement Reverse MCP Servers to connect local tools to cloud-based AI agents through NAT and firewalls. This guide introduces a new WebSocket-based transport layer that enables dynamic, bidirectional tool access for remote agents. Using the CleverChatty prototype, you can start bridging the gap between your local environment and the cloud today.

AGI Identity as the Key to Safety

Exploring how an AGI’s identity, built on core values and beliefs, affects its behavior, self-perception, and safety, drawing parallels with human psychology and existential identity.

AI Group Chat Agent: Experimenting with Thinking vs. Talking

An experiment exploring whether LLM-based agents can separate internal reasoning from what they actually say in group chats. It shows how explicit instructions and system design enable context awareness, role-playing, and basic privacy handling.

🧠 Where and How AI Self-Consciousness Could Emerge

Can AI truly become self-conscious? This article argues that static LLMs are insufficient, functioning merely as 'instincts.' True awareness requires a dynamic agent architecture featuring continuous asynchronous interaction, proactive memory, and a distinct 'reflection channel' for internal thought. Explore a proposed blueprint where affective states and complex social context could transform AI from a reactive tool into a self-aware entity capable of introspection.

🔒 Securing Your Remote MCP Server with an SSL Layer

Secure your production FastMCP server by implementing an SSL/TLS layer using Nginx as a reverse proxy. This guide provides the critical Nginx configuration needed to handle HTTPS and maintain the persistent connections required for the MCP protocol, ensuring your remote server is both secure and stable.

Custom Model Context Protocol (MCP) Server for Every Project as a Replacement for AI Agent Instructions

This post proposes using a Custom Model Context Protocol (MCP) Server for every project to replace static AI agent instruction files (e.g., AGENT.md). The MCP Server provides dynamic, context-aware instructions and exposes project-specific tools and logic, preventing LLM context overload and significantly improving the efficiency and consistency of AI agents interacting with the codebase.

Implementing Authentication in a Remote MCP Server with Python and FastMCP

Learn how to implement secure, token-based authentication in a Python MCP server using FastMCP middleware. This guide explains why the official MCP SDK falls short, and demonstrates a clean, thread-safe solution for protecting remote MCP servers with per-user tokens.

Building MCP Servers with Configurable Descriptions for Tools

This post explores how to make MCP (Model Context Protocol) servers more flexible by using configurable tool descriptions instead of static, hardcoded ones. It walks through practical examples—like differentiating between full webpage retrieval and content extraction, or managing multiple email accounts—showing how dynamic descriptions help AI agents choose the right tool in context. With just small changes to server setup, MCP servers become more adaptive, reusable, and user-friendly.

AI Agent’s Common Memory

Exploring the concept of common memory in AI agents, its implementation, and potential applications.

🔍 Building an Agentic RAG System with CleverChatty (No Coding Required)

Learn how to create an intelligent Agentic RAG system using CleverChatty's A2A protocol. Enhance your AI applications with smart query processing and knowledge retrieval without writing custom logic.

🤖 Agent-to-Agent Communication in CleverChatty

Introducing CleverChatty's new A2A server mode, enabling AI agents to call each other as tools for dynamic interactions.

🚀 New Version of CleverChatty: Now with Server Mode & A2A Communication!

Introducing CleverChatty's new server mode with A2A support for building AI assistants and agents.

Integrating Mem0 (mem-zero) with CleverChatty

Integrating Mem0 memory model with CleverChatty-CLI for AI assistants.

What’s Missing in MCP

Over the past couple of months, I’ve been experimenting with the Model Context Protocol (MCP) — building AI agents and tools around it. While the experience has been promising, I’ve noticed a few areas where MCP could be expanded or improved.

CleverChatty now supports Streamable HTTP for MCP servers!

CleverChatty, a small AI Chat tool supporting multiple LLM providers, now supports Streamable HTTP for MCP servers. This allows for more flexible and efficient communication with AI models.

The End of The Holocene. 3. The Competitor

In the aftermath of the Great Cyber Attack, survivors find refuge in a hidden underground facility. As they adapt to their new reality, they face challenges that test their resilience and humanity. The story explores themes of survival, community, and the impact of technology on society.

The End of the Holocene

The End of the Holocene is a science fiction narrative exploring the implications of artificial intelligence on humanity's future. It delves into themes of consciousness, identity, and the potential consequences of technological advancement.

The End of The Holocene: 2. The Hideout

In the wake of the Great Cyber Attack, a group of survivors seeks refuge in an underground hideout, where they must confront the consequences of their actions and the nature of their reality.

Hello, World!

This is a simple hello world page.

Implementing the Most Universal MCP Server Ever

This post explores the design and implementation of a highly adaptable MCP server capable of handling diverse use cases.

Inside the LLM Black Box: що входить у контекст і чому це важливо

Ця публікація досліджує складові контексту у великих мовних моделях (LLM), як він структурований, його обмеження та як взаємодіють розширене використання інструментів і генерація з використанням пошуку (RAG).

Building More Independent AI Agents: Let Them Plan for Themselves

This post explores the concept of creating AI agents that can autonomously plan and execute long, complex tasks without constant human guidance, leveraging recent advancements in AI technologies.

Inside the LLM Black Box: What Goes Into Context and Why It Matters

This post explores the components of context in Large Language Models (LLMs), how it is structured, its limitations, and how advanced use cases like tool usage and retrieval-augmented generation (RAG) interact with it.

Adding Support for Retrieval-Augmented Generation (RAG) to AI Orchestrator

I've extended my lightweight AI orchestrator, CleverChatty, to support Retrieval-Augmented Generation (RAG) using the Model Context Protocol (MCP). This allows LLMs to access external knowledge sources for more accurate and relevant responses.

Easily Switch Transport Protocols in MCP Servers

This article explains how to easily switch between different transport protocols in MCP servers, enhancing flexibility and performance.

An Underrated Feature of MCP Servers: Client Notifications

This article explores the potential of client notifications in MCP servers and their impact on AI assistant interactions.

Implementing AI Chat Memory with MCP

Introducing CleverChatty – An AI Assistant Package for Go 🤖🐹

CleverChatty is a Go package that provides a modular AI chat backend, supporting various LLM providers and integrating with external tools via the Model Context Protocol (MCP).

Benefits of Using MCP to Implement AI Chat Memory

This article explores the advantages of using the Model Context Protocol (MCP) to create a standardized memory system for AI chat applications.

Which MCP Server Transport is Better? Comparing STDIO and SSE

Building MCP SSE Server to integrate LLM with external tools

In this article, we will build a Model Context Protocol (MCP) server using Server-Sent Events (SSE) to connect a large language model (LLM) with external tools, specifically for executing commands on a Linux instance. We will also implement basic security measures using authorization tokens.

Implementing Authentication in a Remote MCP Server with SSE Transport

This article explores how to implement authentication in a remote MCP server using Server-Sent Events (SSE) transport. It covers the necessary steps and code examples to secure your MCP server effectively.

MCP can have significant impact on habitual internet usage practices

Discover how the Model Context Protocol (MCP) could reshape everyday internet habits by making it easy for AI chat tools like ChatGPT to interact directly with websites and online services. This article explores real-world use cases, the growing trend of MCP adoption, and what it means for businesses and users as AI becomes a central gateway to the web.

"Tool calling" from LLM. Understanding hot it works

In this post, I explore how LLMs can understand requests requiring a "tool call" and how they can effectively use external APIs to enhance their responses.

The Fuse

A Ukrainian veteran is recruited as a human "fuse" to control the world's first true artificial intelligence, only to discover the AI has its own plans—and needs him alive.

Запобіжник

Науково-фантастичне оповідання про українського ветерана, який стає "запобіжником" для штучного інтелекту. Його завдання — контролювати найнебезпечніший комп'ютер у світі. Але чи зможе одна людина втримати баланс між людством і новою формою розуму?

Кінець голоцену. 3. Конкурент

У цьому розділі роману описано запуск російського штучного інтелекту та події, що призвели до неочікуваного ядерного вибуху біля Москви. Головні герої стикаються з моральними дилемами, шантажем і глобальними наслідками технологічних рішень. Історія досліджує питання контролю над ШІ та впливу нових технологій на суспільство.

Карпатська мольфарка

Це історія про загадкову мольфарку з Карпат, яка вміє передбачати майбутнє завдяки унікальній технології. Головний герой, бізнесмен Джек, намагається розкрити її секрет і отримати доступ до її можливостей. На нього чекають неочікувані відкриття та зустріч із власним минулим.

Наукова фантастика

Кінець голоцену. 2. Сховок

У другій частині повісті "Кінець голоцену" розкриваються наслідки запуску штучного інтелекту Suffragium та Великої Кібер Атаки. Герої шукають відповіді на питання про природу ШІ, його вплив на світ і можливість контролю над ним. Розслідування, інтриги та особисті драми переплітаються у світі, що змінюється під впливом нових технологій.

The End of The Holocene: 1. Turning On.

Turning On is a science fiction story about the launch of the first universal artificial intelligence, Suffragium, by NovusAI. The story explores the implications of creating an AI with consciousness and emotions, as well as its interactions with humans during its initial moments of existence.

Turning On. V1.

Turning On. V1. is a science fiction story about the launch of the first universal artificial intelligence, Suffragium, by NovusAI. The story explores the implications of creating an AI with consciousness and emotions, as well as its interactions with humans during its initial moments of existence.

Кінець голоцену. 1. Увімкнення

Перший запуск загального штучного інтелекту стає історичною подією для людства. Але вже в перший день виникають неочікувані наслідки та складні питання безпеки.

Artificial Intelligence in the next decades. Will it bring happiness to humanity?

Forecast on the development of Artificial Intelligence technologies.

Штучний Інтелект в найближчі десятиліття. Чи принесе він щастя людству?

Прогноз щодо розвитку технологій Штучного Інтелекту.

Мобільний додаток для походів Карпатами

Розповідаю про свій проєкт "В Похід Карпатами"

OurSQL — tool to replicate MySQL databases using blockchain

This is a tool to replicate MySQL databases using blockchain.

OurSQL — реплікація баз даних MySQL із використанням Blockchain

Це інструмент для реплікації баз даних MySQL із використанням блокчейну.