AI Content Engine
The PRD's name for AI-driven content generation, enhancement, and analysis across a CMS; in this repo the same goal is achieved through claude-obsidian's skills rather than a dedicated backend service.
Estimated reading time: 2 minutes
Page Improvements
- Improve Page Polish content, structure, and presentation of this page
- Expand Page Add depth, examples, and missing sections to this page
- Update Page Refresh outdated content, versions, links, and screenshots
- Fix Page Issue Report a typo, broken link, layout glitch, or content bug
- SEO Optimize Improve discoverability, metadata, and structured data
- Accessibility Audit Audit this page for WCAG 2.1 AA compliance
Site Improvements
- UI/UX Improvement Propose a design or UX refinement for the theme
- New Feature Propose a new site-wide feature or capability
- Component Enhancement Improve a Jekyll layout, include, or shared component
- Performance Optimization Improve load time, Core Web Vitals, and asset delivery
Table of Contents
AI Content Engine
Definition
The AI Content Engine is the zer0-pages PRD’s (§5.1) name for a subsystem that generates and enhances content using large language models: blog posts from prompts, documentation from repository analysis, PRDs from repo scans, code-quality analysis, and content enhancement, with multi-provider fallback across OpenAI, Anthropic, and xAI models.
How It Works
Per the PRD, the engine is a Django-side service exposed via REST endpoints (/api/ai/generate/, /api/ai/enhance/, /api/ai/analyze/, /api/ai/chat/), with retry/backoff, model fallback on quota exhaustion, response caching, and per-provider timeout handling. This is aspirational — no such API exists in this repository. The functional equivalent that actually exists here is claude-obsidian’s skill set: /wiki-ingest turns a dropped source into structured notes, /save captures a conversation as a note, /autoresearch runs a web-research loop and files its findings, and /wiki-query answers questions from the vault. These are invoked conversationally inside Claude Code, not called as a REST API, and they operate on Markdown files in an Obsidian vault rather than database rows.
Why It Matters
The gap between the PRD’s imagined engine and claude-obsidian’s actual skills is the central “vision vs. reality” tension of this ingestion: the PRD describes a productized, API-driven AI backend; what this repo actually runs is an agent (Claude Code + claude-obsidian) operating directly on a file-based vault that is simultaneously the Jekyll site source. Understanding this distinction matters for anyone reading the PRD literally and expecting to find a /api/ai/generate/ endpoint, a Django admin, or a hosted AI service in this codebase — none exists.
Examples
- PRD’s imagined capability: “PRD Generation: Analyze GitHub repos and produce comprehensive PRDs” — see PRD Machine for the PRD’s own (also unimplemented) framing of this.
- Real capability in this repo: this very page was produced by exactly the kind of process the PRD describes (AI reads a source, extracts entities/concepts, files structured notes) — but performed by claude-obsidian’s
/wiki-ingestskill acting on Markdown, not a Django “AI Layer” acting on a database.
Connections
- claude-obsidian — the real implementation of “AI-generates-content-into-the-vault.”
- zer0-pages — the product name under which the PRD’s vision (and the repo’s reality) both live.
- PRD Machine — a narrower, PRD-specific instance of the same “AI generates structured documents” idea.
- Jekyll/Obsidian Dual-Rendering Architecture — the actual substrate the AI-authored content is written into and published from.
Sources
- zer0-pages PRD, §5.1 “AI Content Engine.”