Cole Medin founded Dynamous, an AI education community, and publishes practical agent-development projects as coleam00. His tutorials cover context for coding agents, retrieval from documents and running AI tools locally.[1][2][3]
Start with Context Engineering Intro →
Career
Medin’s public work combines a YouTube channel, open-source repositories and the Dynamous education community. His conference biography identifies him as Dynamous founder and CEO and describes work on custom agents, retrieval systems and local AI deployment. The common format is hands-on instruction: a learner can inspect or run a project rather than relying only on an abstract description of an agent.[1][2][3]
Contributions
Giving a coding agent the missing information
Context Engineering Intro provides a project template with rules, examples, documentation and a feature request. Its workflow first generates a Product Requirements Prompt, an implementation plan tailored to the repository, then executes the plan with validation. The important idea is that instructions alone may omit the surrounding code patterns and checks. Supplying that context makes the desired behavior more explicit and gives the agent a way to tell whether its change succeeded. The repository’s claimed multipliers are promotional comparisons, not independently measured results.[4]
Putting local AI components together
Local AI Packaged groups tools such as Ollama, n8n and Open WebUI into a deployable local setup. Those components serve different jobs: model execution, workflow automation and a chat interface. Packaging them together gives learners a concrete environment in which to connect an application to a model and supporting services. It is integration and teaching work, with the underlying software still credited to its own projects.[5]
Tools, writing & talks
Writing, research & talks
Start here
Context Engineering Intro
github.comA template for giving a coding agent repository rules, examples, feature requirements and validation steps.
Start with Context Engineering Intro if you already use a coding assistant. Adapt its rules and examples to a small existing project, then examine the validation steps before trying the larger local-AI environment.
Repository owner and educator[1][4]Original source- AI Agents Masterclassgithub.com
Cole Medin's hands-on AI agents masterclass — video series with companion code covering production-grade agent patterns.
Catalog workOriginal source
Tools
- Midnight Conversegithub.com
Experimental late-night AI conversation app built with TypeScript, Vite, and Tailwind.
Catalog workOriginal source - GitHub Issue Triagergithub.com
Issue triage dashboard on Next.js and pgvector that classifies, embeds, and plans implementation for every repo issue.
Catalog workOriginal source - Helplinegithub.com
Demonstration codebase for building an AI layer in large projects with CLAUDE.md hierarchy, hooks, skills, LSP, and MCP.
Catalog workOriginal source - Obsidian Voice Agentgithub.com
Voice-powered Obsidian agent for interacting with a knowledge base through speech.
Catalog workOriginal source - Evolution of AI Agentsgithub.com
Three working examples tracing how agent development moved from traditional RAG to SDKs and skill-based frameworks.
Catalog workOriginal source - Codex Telegram Coding Assistantgithub.com
Codex-powered assistant wired into Telegram so coding requests can be sent from a phone.
Catalog workOriginal source - Arcade AI Agentgithub.com
Python agent that reaches into Gmail and Asana through Arcade to carry out tasks for the account owner.
Catalog workOriginal source - HyperFrames AI Video Generationgithub.com
Archon workflow that takes any URL and produces a finished AI-voiced HyperFrames short video from it.
Catalog workOriginal source - Dynamous AI Coachgithub.com
Python RAG agent that coaches users with Cole's content as its knowledge base.
Catalog workOriginal source - Agentic Chat Application Templategithub.com
Scaffold for agentic chat apps, shipping with Next.js, TypeScript, and shadcn/ui already wired up.
Catalog workOriginal source - Human-in-the-Loop RAG Agentgithub.com
RAG agent built with CopilotKit and Pydantic AI that lets a human validate retrieved sources in real time.
Catalog workOriginal source - Custom Agent with Skillsgithub.com
Pydantic AI agent that reproduces Claude Skills-style progressive disclosure without depending on Claude.
Catalog workOriginal source - Dark Factory Experimentgithub.com
AI chat application that answers questions over YouTube video transcripts using retrieval-augmented generation.
Catalog workOriginal source - AI Transformation Workshopgithub.com
Workshop materials on agentic engineering covering the AI layer concept, the PIV loop, and 15 Claude Code commands.
Catalog workOriginal source - Adversarial Devgithub.com
Three-agent harness that pits a generator against an adversarial evaluator to build applications with quality gates.
Catalog workOriginal source - MongoDB RAG Agentgithub.com
Hybrid RAG agent on MongoDB and Pydantic AI that fuses semantic and text search with reciprocal rank fusion.
Catalog workOriginal source - Link-in-Bio Page Builderlink-in-bio-page-builder.vercel.app
Open-source Next.js builder for Linktree-style link pages you can self-host on Vercel.
Catalog workOriginal source - Supabase MCPgithub.com
Server that lets AI agents query, mutate, and manage Supabase projects through the Model Context Protocol.
Catalog workOriginal source - Your Claude Engineergithub.com
Agent harness with Slack, GitHub, and Linear tools that turns Claude into an end-to-end software engineer.
Catalog workOriginal source - PydanticAI Research Agentgithub.com
Research agent built with Pydantic AI that follows the PRP framework template for production use.
Catalog workOriginal source - Linear Coding Agent Harnessgithub.com
Harness that turns a Linear issue into an assignment for an autonomous coding agent, from code to pull request.
Catalog workOriginal source - Ralph Loop Quickstartgithub.com
Shell quickstart for running the Ralph Wiggum agent loop without depending on the Anthropic plugin.
Catalog workOriginal source - Remote MCP Server with Authgithub.com
Starter template for building a remote MCP server with GitHub OAuth authentication scaffolding.
Catalog workOriginal source - Remote Agentic Coding Systemgithub.com
Wires Claude Code or Codex into Slack, Telegram, and GitHub so coding work can be driven remotely.
Catalog workOriginal source - Second Brain Startergithub.com
Claude Code skill that generates a personalized PRD for building a proactive, persistent AI assistant.
Catalog workOriginal source - MCP Mem0github.com
MCP server that gives agents long-term memory via Mem0, doubling as a template for Python MCP servers.
Catalog workOriginal source - Second Brain Skillsgithub.com
Collection of Claude skills that turns Claude Code into a second-brain workspace for knowledge capture and retrieval.
Catalog workOriginal source - Claude Memory Compilergithub.com
Persistent memory for Claude Code that captures sessions via hooks and compiles decisions into organized notes.
Catalog workOriginal source - MCP Crawl4AI RAGgithub.com
MCP server for web crawling and RAG that lets AI coding assistants fetch, parse, and index live documentation.
Catalog workOriginal source - Excalidraw Diagram Skillgithub.com
Claude Code skill that lets a coding agent generate practical Excalidraw diagrams from a text description.
Catalog workOriginal source - Local AI Packagedgithub.com
A local AI environment combining model execution, workflows and supporting interfaces.
Repository owner and integrator[1][5]Original source - oTTomator Agentsgithub.com
Open-source collection of production-grade AI agent reference implementations from the oTTomator Live Agent Studio.
Catalog workOriginal source