GitHub Pages VitePress English Tieng Viet Mermaid License: MIT

AI Engineering Stack Guide social preview

A bilingual field guide and practical toolkit for understanding the modern AI engineering stack: AI-DLC, Spec-Driven Development, workflow frameworks, agent harnesses/runtimes, agent app frameworks, model serving, RAG/data, MCP/tools, evals, observability, security, and governance.

Live site: https://anhtnt90dev.github.io/ai-engineering-stack-guide/

AI engineering tools are increasingly hard to compare because many of them use the same verbs:

plan -> implement -> review -> iterate

That similarity creates confusion. Spec Kit, OpenSpec, AWS AI-DLC, GSD, Superpowers, Hermes, Codex CLI, Claude Code, LangChain, LangGraph, and MCP can all appear in an AI-assisted engineering workflow, but they do not solve the same problem.

This guide explains the difference by layer:

Model / Serving
 -> Data / RAG
 -> Tools / MCP
 -> Agent App Frameworks
 -> Agent Harnesses / Runtimes
 -> Workflow / Methodology
 -> Artifacts / Source of Truth
 -> Evals / Observability / Governance

The goal is to help readers choose the right tool for the right layer instead of comparing unrelated frameworks as if they were direct competitors.

• Why AI-DLC exists and how it changes traditional software delivery.
• What Spec-Driven Development means in AI-assisted coding.
• How GitHub Spec Kit, OpenSpec, AWS AI-DLC Workflows, GSD, and Superpowers differ.
• Where Hermes, Codex CLI, and Claude Code fit as agent harness/runtime tools.
• Where LangChain and LangGraph fit as agent app frameworks.
• Why RAG, MCP/tools, evals, observability, security, and governance are separate production layers.
• How to combine frameworks without creating multiple sources of truth.
• Which stack fits common use cases such as SaaS features, RAG products, enterprise modernization, internal agent platforms, and long-running agent services.
• How to use a one-page cheat sheet, interactive decision wizard, templates, and scenario lab to apply the concepts in real projects.

Recommended path:

1. Start with the Stack Map.
2. Use the Cheat Sheet and Decision Wizard.
3. Download the Templates and Starter Artifacts.
4. Read AI-DLC and Spec-Driven Development foundations.
5. Understand Agent Harness vs Workflow Framework.
6. Read the deep dives for each workflow framework.
7. Read LangChain, LangGraph, and Hermes positioning.
8. Use the comparison matrix, scenario lab, and decision guide.
9. Apply the reference architectures and adoption playbook.

• GitHub Spec Kit: https://github.com/github/spec-kit
• OpenSpec: https://github.com/Fission-AI/OpenSpec
• AWS AI-DLC Workflows: https://github.com/awslabs/aidlc-workflows
• GSD Core: https://github.com/open-gsd/gsd-core
• Superpowers: https://github.com/obra/superpowers
• Hermes Agent: https://github.com/NousResearch/hermes-agent
• LangChain: https://docs.langchain.com/oss/python/langchain/overview
• LangGraph: https://docs.langchain.com/oss/python/langgraph/overview
• OpenAI Agents SDK: https://platform.openai.com/docs/guides/agents-sdk/
• Microsoft AutoGen: https://microsoft.github.io/autogen/
• CrewAI: https://docs.crewai.com/
• Google Agent Development Kit: https://google.github.io/adk-docs/
• Azure AI Foundry Agent Service: https://learn.microsoft.com/azure/ai-foundry/agents/overview
• Amazon Bedrock Agents: https://docs.aws.amazon.com/bedrock/latest/userguide/agents.html
• Dify: https://docs.dify.ai/
• n8n AI Agent node: https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent/
• Model Context Protocol: https://modelcontextprotocol.io/
• OpenTelemetry: https://opentelemetry.io/

Requirements:

• Node.js 22 or newer is recommended.
• npm.

Install dependencies:

npm install

Run the local documentation server:

npm run docs:dev

Build the static site:

npm run docs:build

Preview the production build:

npm run docs:preview

The English site is under /en/. The Vietnamese site is under /vi/.

This repository uses GitHub Actions to deploy VitePress to GitHub Pages.

Workflow file:

.github/workflows/deploy.yml

The workflow sets:

BASE_PATH=/${{ github.event.repository.name }}/

That makes the site work as a GitHub project page:

https://anhtnt90dev.github.io/ai-engineering-stack-guide/

docs/
 .vitepress/
 config.mts
 theme/
 en/
 foundations/
 tools/
 stack/
 frameworks/
 app-frameworks/
 harnesses/
 compare/
 vi/
 foundations/
 tools/
 stack/
 frameworks/
 app-frameworks/
 harnesses/
 compare/
 public/
 templates/

This is an evolving learning guide and practical toolkit. It currently includes:

• Bilingual English and Vietnamese documentation.
• Layer-first taxonomy for AI workflows, harnesses, app frameworks, RAG, tools, evals, observability, security, and governance.
• Deep dives for Spec Kit, OpenSpec, AWS AI-DLC Workflows, GSD, Superpowers, Hermes, LangChain, and LangGraph.
• A one-page cheat sheet and interactive decision wizard.
• Downloadable templates for specs, AI-DLC records, GSD plans, TDD prompts, LangGraph state design, RAG evals, tool permissions, and adoption scoring.
• A scenario lab showing the same RAG support assistant through multiple workflow lenses.
• An adjacent ecosystem map for OpenAI Agents SDK, AutoGen, CrewAI, Google ADK, Azure AI Foundry Agents, Amazon Bedrock Agents, Dify, n8n, LlamaIndex, Haystack, and Semantic Kernel.

MIT. See LICENSE.