Changelog

Theme-safe visuals and editorial cover system

FullStack AI Hub now combines accessible Knowledge Map labels, a keyboard-scrollable release timeline, and a responsive generated cover system across its core learning surfaces.

• Replaced hard-coded map label colors with semantic theme surfaces and foregrounds.
• Made the homepage changelog independently scrollable with keyboard focus and a visible scrollbar.
• Added light and dark visual regression coverage for Knowledge Map labels.
• Added optimized, text-free editorial covers for the production stack, guided learning, architecture guides, and projects.

Mapped concepts

207

Complete lessons

75

Labs

14

Projects

6

Assessments

75

Lab scenarios

42

Guides

8

System inspectors

2

Learning evidence and applied practice

FullStack AI Hub now connects researched lessons, assessed projects, mastery, review scheduling, production labs, and reference architectures into one evidence-based learning system.

• Expanded the public curriculum to 75 researched lessons with deterministic assessment packs.
• Added multidimensional mastery, adaptive reviews, explainable recommendations, and Progress V4 portfolio export.
• Upgraded all six projects with weighted rubrics, portable evidence, completion reports, and two capstones.
• Added 42 scenarios across 14 labs and published eight production reference architecture guides.
• Rebuilt the Codex and vLLM inspectors around immutable versions and classified evidence.

Project-based production AI learning

Six connected engineering projects now turn architecture decisions, system design, reliability, evaluation, and optimization work into portable learning evidence.

• Added production RAG and governed agent capstones.
• Connected milestones to lessons, labs, paths, and architecture guides.

Fourteen deterministic engineering labs

The lab catalog now covers retrieval, agents, evaluation, inference, observability, security, cost, latency, tokenization, and decoding decisions.

• Added beginner, intermediate, and production modes to every lab.
• Added saved attempts, comparisons, failure injection, explanations, and result exports.

Evidence-audited Codex and vLLM inspectors

System diagrams now separate verified implementation evidence from explicit inference and conceptual teaching models.

• Pinned both breakdowns to immutable upstream commits and releases.
• Added claim-level sources, evidence labels, review dates, and inference rationale.

FullStack AI Hub moved to its Next.js learning platform

The curriculum, Knowledge Map, learning workspace, labs, projects, guides, bookmarks, and local progress now run as one independent production application.

• Replaced the previous publishing surface with a purpose-built learning workspace.
• Added direct VPS deployment, health checks, rollback releases, and privacy-audited Docker contexts.