Skip to main content

ChatMCP: Conversational AI

chatMCP is a mobile-first messaging app inspired by WhatsApp, designed for seamless communication between humans and AI agents via the Model Context Protocol (MCP). It treats MCP servers as contacts, supports group chats with both humans and AI agents, and offers a clean, intuitive UI optimized for mobile devices.

🚀 Features​

  • MCP Server Contacts: Add AI agents by MCP server URL or identifier; add humans by phone, email, or username.
  • Group Chats: Mix humans and AI agents in group conversations.
  • Messaging: Send/receive text, images, files, and voice messages.
  • Search: Quickly find contacts and messages.
  • Guest Access: Interact with AI agents without registration; upgrade to a full account via phone/email verification.
  • OAuth Integration: Secure authentication with MCP servers.

Designing conversational flows in ChatMCP means orchestrating how users, agents, and context interact through structured APIs. By leveraging OpenAPI and the Model Context Protocol (MCP), you can create scalable, maintainable, and context-aware chat systems.

Context Engineering in chatMCP​

ChatMCP uses Context Engineering to orchestrate the flow of information between user input, prompt construction, agent reasoning, retrieval-augmented generation (RAG), tool invocation, and memory management.

ThechatMCPContext EngineeringUserInput1Prompt5🤖 Agent2🔍 RAGvector search3🧩 MCP Tools</>📄⏰🔧4Long-Term MemoryMCP Servers🎛️🔄😁☁️🧩Databases📊📈🗄️💾Add tomemory8Short-Term Memory7store all context inchat historyAnswerHAPI formatted response6update promptwith answeragenticcoordination &decision makingHAPIEcosystemPart of The Happiness StackLegendUser InputRAG ContextAgent / AnswerAction ToolsUser InfoChat History

Key Flow Steps:

  1. User Input: The user sends a message, which is added to the prompt and chat history.
  2. Prompt Construction: The prompt is built from user input and short-term memory (chat history).
  3. Agent Reasoning: The agent processes the prompt, coordinating with RAG and tools as needed.
  4. RAG Context: The agent may query long-term memory (vector search) for relevant context.
  5. Action Tools: The agent invokes tools (OpenAPI operations) for structured actions.
  6. Answer Generation: The agent produces an answer, which is returned to the user and added to memory.
  7. Memory Management: All context is stored in short-term (chat history) and optionally long-term memory (MCP servers/databases).
tip

Context Engineering ensures every chat flow is context-aware, API-first, and memory-driven.