A growing ecosystem of hands-on guides, runnable code, full tutorials, and reference examples — RAG pipelines, multi-agent crews, MCP tooling, and production-style APIs. Every system built local-first and private before a single cloud key gets pasted.
Maintained by Ayush Kumar and contributors. Run on your own machine.
How this works
Most agent material teaches in scattered pieces. A RAG demo here, a fine-tuning post there, a flashy multi-agent screenshot somewhere else. The pieces rarely line up — you ship a chatbot but can't trace its retrieval, or you wire a tool to an agent and can't say what the harness around it is actually doing.
This repo is the spine. Seventeen self-contained guides, each a real project under guides/<name>/, from a two-agent RAG API to a self-hosted assistant to a full agents masterclass.
Every guide runs the same loop: read the problem, draw the architecture, write the code, run it locally, keep the artifact. No five-minute videos, no copy-paste cloud deploys, no hand-holding.
Free, open source, and built to run on your own laptop. Clone the repo, pick a guide, and go from idea to a running local agent in an afternoon — the site mirrors every README and tutorial so you can read first and clone later.
Catalog
Agent anatomy, ReAct/ReWOO, 15+ frameworks, use cases, MCP/A2A, and five runnable code examples.
→ OpenComplete Hermes guide — learning loop, Profile Builder, memory, Curator, GEPA, and three agents.
→ OpenPrivate two-agent RAG API with Qwen (Ollama), CrewAI, Qdrant, Firecrawl, LitServe, and Gradio.
→ OpenMessaging assistant on gemma4:e2b with a local RAG skill calling the LitServe API.
→ OpenTeam-aware Claude Code layout — CLAUDE.md, permissions, rules, skills, and subagents.
→ OpenInstall and extend the self-hosted personal AI assistant — gateway, channels, skills, cron.
→ OpenInstall Hermes Agent and map the skills, plugins, GUIs, and integrations ecosystem.
→ OpenCompare Hermes Agent and OpenClaw — gateways, skills, migration, and a decision guide.
→ OpenModel Context Protocol — host/client/server, capability exchange, API vs MCP, App MCP.
→ OpenVisual ML pipeline — problem framing through deploy and monitor, with animated GIFs.
→ OpenBuild PulseFeedback with 10 equal tools — PM to support in one closed loop.
→ OpenAct, observe, repeat — eval gates, open vs closed loops, fleet patterns, ReAct.
→ OpenAgent = Model + Harness — five subsystems, session lifecycle, verification gates.
→ OpenLoRA, QLoRA, RLHF, DPO, GRPO — when to fine-tune, PEFT math, HuggingFace walkthroughs.
→ OpenQuick index — the full Profile Builder walkthrough lives in Masterclass Parts 11–12.
→ Open754 MITRE-mapped security skills for AI agents — install, frameworks, SOC walkthroughs.
→ OpenGive your agent a safe credit card — provision/list/rotate/teardown infra over one MCP protocol, with Supermemory as the headline provider.
→ OpenNew guides, runnable code labs, live demos, and video walkthroughs are on the way. Star the repo to follow along.
→ Watch on GitHubColophon
The entire ecosystem is on GitHub. Clone it, fork it, learn at your own pace — no paywall, no signup. Every guide ships its own requirements.txt and .env.example.
# grab the repo and pick any guidegit clone https://github.com/Ayush7614/agentic-ai-ecosystem.gitcd agentic-ai-ecosystem/guides/<guide-name> # isolated environment python -m venv .venv && source .venv/bin/activate pip install -r requirements.txt cp .env.example .env