Docs

The architecture for content systems that serve humans, agents, and LLMs equally well: pure Markdown, deterministic tools, clear boundaries.

A practical architecture for human docs and agentic systems: pure Markdown for meaning, MCP/tools for action.

The practical architecture: Markdown for durable meaning, TSX for behavior, MCP/tools for action. MDX is optional and often unnecessary.

A practical operating plan for building GSC after clarifying the Markdown, TSX, MDX, and MCP boundaries.

Architecture and UX plan for a standalone CLI tool that scaffolds new projects from the boilerplate via a web UI wizard.

My next evolution for AI-assisted content operations: generate structured documents with rich metadata, edit them quickly in a dedicated review surface, and publish high-quality artifacts in under a minute.

Debugging notes on building a bookmark that opens two URLs simultaneously in separate tabs — what Chrome blocks, what works, and why.

I wired up page title metadata across the whole site — except for three pages that use 'use client'. I know the fix. I don't want to do it yet. Here's everything documented so I can come back to it.

Implementation plan for building a GitHub Copilot skill that enables natural language content creation

A development-only inline content editor for markdown files that enables quick content updates without manually editing files in an IDE.

Reference doc for the SQLite migration. Captures exactly how the current filesystem-based content system works, where the pain points are, and what each function actually does.

How context awareness, reasoning between tool calls, and adaptive execution make the Copilot SDK fundamentally different from traditional automation

Test page for every configured CodeBlock language rendered through the markdown pipeline.

Stop building separate tools - use Next.js /dev/* routes to turn your entire app into a personal productivity layer with hot reload, TypeScript safety, and zero deployment overhead

A frictionless life logging and time tracking system using natural language AI interface for effortless tracking and behavioral insights

Spec for a /live route — a personal microblog feed powered by a write API and SQLite, postable from anywhere including a phone.

Design document for an AI-powered component composition system using LLM intelligence with deterministic, type-safe output

A practical guide to optimizing your site for AI-powered search and LLM discoverability - actionable strategies for appearing in ChatGPT, Claude, Perplexity, and other LLM-generated responses

Testing GFM table syntax rendering with primitive table components.

TSX-first content system with markdown fallback: embracing coupled content and logic

A focused workflow for capturing ideas into backlog storage when I intentionally want to stop implementing and preserve context for later.

Frictionless system for showcasing development projects - make adding new projects feel as easy as publishing a blog post

Replacing Fuse.js title-only search with server-side full-text search using SQLite FTS5 — paragraph-level results with highlighted snippet previews linking directly to headings.

A taskbar-style bottom navigation pattern that keeps search accessible in the mobile thumb zone and provides app-like ergonomics.

Add visual thumbnail support to all content types (posts, projects, pages) for improved visual appeal, social sharing, and user engagement

Security design notes for preventing prompt-injection abuse in a natural-language time-blocking app via AI intent validation.

Figure out whether to create dedicated /docs page or surface all docs on homepage via Navigator

Add optional metadata field to hide ContentHeader component for non-standard pages like about-me

Design patterns and status tracking improvements for managing todos-as-content-pages.

Action plan to standardize authorship metadata and workflow transparency across all posts.

A unified content model where every item is a source plus commentary, rendered differently based on type.

Architecture plan for migrating a pool dashboard from localStorage to Turso with passphrase-protected writes and public reads.

The full technology stack behind jaygriff.com — what's used, why it was chosen, and how it all fits together.

Why sidebar logos get compressed while icons stay perfect: a CSS selector mismatch between <svg> and <img> elements.

Open and completed tasks for website development and bug fixes

A gamified system for tracking your progress through your favorite creators' complete bodies of work.

Complete annotated file tree documentation for the jaygriff-v2 project architecture

Two-phase plan for building a simple admin interface with password auth now and passkey upgrade later.

Why chunking Markdown into structured blocks unlocks vector search, WYSIWYG editing, and inline search results.

Introducing Projects as a first-class organizing concept to give docs, posts, and demos the context they need.

How styling works in this project — CSS modules, cn(), theme variables, and the rules that keep it consistent.

Technical reference for the content pipeline covering Turso schema, sync script, AI metadata generator, markdown renderer, and route architecture.

Building a local admin interface with passwordless WebAuthn authentication using Bitwarden's passwordless.dev service.

A filesystem-first content architecture with AI-generated metadata, Turso indexing, and dynamic routing.