Main/WordPress Fullstack/Blueprints/AI & Agents
WordPress54 ТЗReact & Next.js10 ТЗAI Engineering6 ТЗ
To all 6 blueprint sections
Custom ThemesGutenberg & FSEWooCommerceDevOps & SpeedSecurity & APIAI & Agents
Раздел • AI & Agents • 5 ТЗ

AI & Agents

5 production specs: CRM MCP Server with RBAC, Self-Healing CI/CD in GitHub Actions, On-Premise Ollama + LiteLLM cluster, Generative UI dashboard, and WooCommerce semantic search.

Section «AI & Agents» Progress
0%(0/25 steps)
+0 XP0 of 5 solved

Foundational Articles & Theory for Zero-to-Pro Learning

6 articles

Master these key 6 materials to execute blueprints in this section with confidence and zero knowledge gaps.

Фундаментальная базаMiddle • Архитектура

WordPress 7 Native AI Client и Abilities API

Нативные интерфейсы ИИ в ядре WordPress 7: интеграция локальных и облачных LLM, генерация метаданных и контента.

~25 minOpen Lesson
Архитектурный паттернSenior • Production

Model Context Protocol (MCP) адаптер для WordPress

Подключение сайта WordPress как источника инструментов и контекста для автономных агентов Claude, Cursor и Gemini.

~30 minOpen Lesson
Архитектурный паттернSenior • Production

Семантический векторный поиск товаров pgvector в WooCommerce

Векторизация каталога товаров, генерация эмбеддингов и поиск по смыслу с гибридным ранжированием BM25.

~30 minOpen Lesson
Архитектурный паттернMiddle • Архитектура

AI-аудит безопасности плагинов и тем WordPress

Автоматизированный поиск уязвимостей, бэкдоров и небезопасных функций ядра через специализированных LLM-агентов.

~25 minOpen Lesson
Практический воркфлоуMiddle • Архитектура

Воркфлоу: Разработка собственного MCP-сервера на TypeScript

Создание сервера Model Context Protocol с валидированными инструментами Zod и подключением к Claude/Cursor.

~40 minOpen Lesson
Практический воркфлоуSenior • Production

Воркфлоу: Enterprise RAG пайплайн на pgvector

Корпоративный RAG-пайплайн: семантическое чанкование, векторный поиск HNSW и Re-ranking через Cohere.

~45 minOpen Lesson
AI & Agents
Senior~12 h

Spec: Production MCP Server for CRM/ERP with RBAC and Rate Limits

Building a production-ready Model Context Protocol (MCP) server in TypeScript for secure integration of autonomous agents with corporate CRM: stdio and SSE transports, Zod validation, RBAC security, audit logging, and PII masking.

“Data security is priority #1. No agent should access deals of other branches or delete a contact without supervisor authorization.”
TypeScript 5.6+@modelcontextprotocol/sdkZod v3Fastify 4 / Express SSE+3
Progress: 0 of 5 steps0%
AI & Agents
Senior~10 h

Spec: Self-Healing CI/CD in GitHub Actions with Automated Bug Fixes

Designing an autonomous Self-Healing pipeline in GitHub Actions: intercepting test failures (Vitest/PHPUnit), extracting stack traces and AST context, calling LLM agents for minimal patch generation, and creating automated Pull Requests.

“The bot must fix the code, NOT weaken or disable the failing unit tests.”
GitHub Actions WorkflowDocker Isolated RunnerVitest / PlaywrightAnthropic Claude 3.7 / Claude Code+2
Progress: 0 of 5 steps0%
AI & Agents
Senior~14 h

Spec: Fault-Tolerant On-Premise Ollama + LiteLLM Cluster with SSO

Deploying an on-premise local LLM gateway with high availability: multi-node Ollama inference cluster (Llama 3.3, Qwen 2.5 Coder), LiteLLM proxy with load balancing and circuit breaker, semantic caching on Redis Stack, and OIDC corporate SSO.

“No single developer query or token may reach public external APIs without explicit compliance approval.”
Docker Compose / SwarmOllama 0.5+ / vLLMLiteLLM Proxy v1.50+Keycloak 25 (OIDC SSO)+3
Progress: 0 of 5 steps0%
AI & Agents
Senior~12 h

Spec: Generative UI Analytics Dashboard on Next.js 15 App Router

Architecting an interactive Generative UI analytics dashboard using Next.js 15, Vercel AI SDK, and Recharts: streaming tool calls, dynamic React component hydration on the fly, widget pin-to-dashboard state persistence, and PDF export.

“We don't want plain text responses with numbers; we want full interactive React visual widgets rendered dynamically.”
Next.js 15+ App RouterReact 19 Server ComponentsVercel AI SDK 4.x (streamText, createDataStreamResponse)Recharts / Tremor UI+3
Progress: 0 of 5 steps0%
AI & Agents
Senior~14 h

Spec: Semantic Vector Product Search with pgvector in WooCommerce (50k SKU)

Engineering a high-speed semantic product search for WooCommerce stores with 50,000+ SKUs: PostgreSQL pgvector database, HNSW indexing, asynchronous embedding generation via Action Scheduler, hybrid search (RRF), and instant AJAX frontend integration.

“Search is the #1 revenue driver. It must be blazing fast (< 150 ms) and handle typos, intent, and synonyms flawlessly.”
WordPress 6.7+ / PHP 8.3WooCommerce 9.0+ HPOSPostgreSQL 16 + pgvectorAction Scheduler (WP Background Tasks)+3
Progress: 0 of 5 steps0%