Skip to main content
Back

Marktechpost's Open Agent Repo Hits 2.9k Stars Covering Every AI Pattern

A sprawling Marktechpost repo pairs 1,100+ commits of runnable notebooks with matching tutorials, covering memory, planning, and multi-agent orchestration end to end.

Repo, Full Access
  • Marktechpost's AI-Agents-Projects-Tutorials repo hit 2.9k stars with 1,117 commits of runnable agent notebooks
  • Covers LangGraph, CrewAI, AutoGen, Google ADK, SmolAgents, PydanticAI, CAMEL, and OpenAI Swarm implementations
  • Deep coverage of memory layers including Mem0, Memori, Cognee, GBrain, and EverMem
  • Includes MCP, A2A, and ACP protocol tutorials for multi-agent communication and tool wiring
  • Recent notebooks focus on skills, governance, policy engines, and self-evolving agent runtimes
  • Every notebook pairs with a companion tutorial writeup, most runnable directly in Google Colab

Learning to build agents from scattered blog posts is painful. Frameworks change weekly, memory patterns are inconsistent, and most tutorials stop at a single tool-calling loop. A growing GitHub collection tries to fix that by shipping runnable notebooks for almost every agent pattern in circulation, paired with matching writeups.

The AI-Agents-Projects-Tutorials repo from Marktechpost has crossed 2.9k stars with 619 forks and 1,117 commits, making it one of the more comprehensive open catalogs of agent implementations available. The tagline is deceptively short: multi-agent systems, memory, planning, reasoning loops. In practice, the repo works as a running reference for how the agent stack has been assembling itself.

What is actually inside

The repository is organized by topic folder, and the breadth reflects how fragmented the agent ecosystem currently is. Folders include A2A Simple Agent, Agentic AI Memory, Agentic Workflows, Agent Communication Protocol, MCP Codes, RAG, Reinforcement Learning, Prompt Optimization, LLM Evaluation, and Voice AI, among many others. Most content ships as Jupyter notebooks that run on Colab, with a smaller number of standalone Python scripts.

The tutorials cut across the ecosystem instead of committing to a single framework. Working examples cover:

  • Orchestration frameworks: LangGraph, CrewAI, AutoGen, Google ADK, CAMEL, SmolAgents, PydanticAI, AgentScope, Griptape, Haystack, LlamaIndex, and OpenAI Swarm
  • Memory layers: Mem0, Memori, Cognee, GBrain, EverMem, and hybrid short-term plus vector long-term memory
  • Protocols: Model Context Protocol (MCP), Agent-to-Agent (A2A), and Agent Communication Protocol (ACP)
  • Model providers: Gemini, Claude, GPT-5, Mistral, Llama 3, Qwen, TinyLlama, and local Hugging Face pipelines