PEPAGI

PEPAGI

**Neuro-Evolutionary eXecution & Unified Synthesis — AGI-like Multi-Agent Orchestration Platform**

A TypeScript multi-agent system where a central **Mediator** (powered by Claude Opus) receives tasks, decomposes them into subtasks, delegates to specialized worker agents (Claude, GPT, Gemini, Ollama, LM Studio), evaluates results, and iterates until the task is complete.

Runs on **Telegram, WhatsApp, Discord, iMessage, CLI**, and **MCP**.

[Dokumentace v cestine / Czech docs](README.cs.md)

---


---

What is PEPAGI?

PEPAGI is an autonomous AI agent orchestrator that acts as a "brain" coordinating multiple AI models. Instead of relying on a single LLM, it intelligently routes tasks to the most suitable model based on difficulty, cost, and capability — then verifies the results using metacognition, world-model simulation, and cross-model verification.

Think of it as an **AI team manager** that knows which AI to ask, when to decompose problems, when to retry with a different strategy, and how to learn from past experience.

---

Key Features

Multi-Agent Orchestration

• Mediator-driven architecture — Claude Opus acts as the central brain, making all routing and decomposition decisions
• Agent Pool — supports Claude (API + CLI OAuth), GPT, Gemini, Ollama, and LM Studio
• Difficulty-aware routing — classifies tasks (trivial → complex) and picks th