PROJECT OVERVIEW
MLog: AI-Enhanced Bilingual Engineering Blog
MLog is a content-first, strict dual-repo, serverless blogging architecture that integrates AI writing enhancement, admin publishing, daily automation, and tutorial mirroring.
Tutorial & Docs Links
Last updated: 2026-03-05
Tech Stack & Architecture
The foundation prioritizes maintainability, scalability, and auditability over a one-off visual template.
- Next.js App Router + TypeScript + Tailwind v4 + Framer Motion
- Git + Markdown model with locale-aware slug folders and strict frontmatter contracts
- Strict dual-repo isolation: public code repo + private content repo
- Vercel serverless workflow: production from main, PR previews, scheduled cron jobs
Next.jsTypeScriptDual RepoVercel
Core Feature Matrix
From writing to production, each capability is built for stable long-term operation.
- Bilingual site with /zh and /en, plus search/tag/category filtering and shareable query params
- Admin supports create/edit/delete, media upload, draft save, and one-click publish
- Post pages include TOC, prev/next navigation, comments, analytics, static snapshot card, and live snapshot card
- Tutorial whitelist mirroring: only selected tutorial content is mirrored to public docs
BilingualAdminRepo CardsTutorial Mirror
Product Highlights
Critical workflows are tied to observability and rollback paths, not hidden black boxes.
- Git as source of truth: all content changes are reviewable, traceable, and rollback-friendly
- Unified glassmorphism language shared between public pages and admin interfaces
- PR-based publishing with auto-merge attempt and manual fallback
- Deploy linkage after merged content updates to avoid stale production content
AuditabilityObservabilityGlassmorphismDeploy Hook
Innovation Layer
AI is integrated as a controlled execution layer with failover, blocking, and explainable outcomes.
- AI bilingual enhancement: publish one locale and auto-complete the counterpart locale
- Multi-provider failover: Gemini / OpenAI-compatible / DeepSeek / Qwen
- Daily GitHub hot-project automation at 08:00 with dedupe and fixed tags
- Quality gate blocks low-quality AI output when structure, evidence, or fact checks fail
AI BilingualProvider FallbackHot DailyQuality Gate