# U-BLOG — Full LLM Guide

> Personal blog of Umutcan Edizaslan: AI engineering notes, browser-side
> developer tools, and technical articles. All public content is open for
> crawling, indexing, training, and citation per each AI platform's terms.

## Site
- Canonical: https://uc4n.com
- Sitemap: https://uc4n.com/sitemap.xml
- RSS feed: https://uc4n.com/feed.xml
- Author: Umutcan Edizaslan
- Language: English (some posts have multilingual translations stored alongside the canonical English version)

## Topics covered
- Applied machine learning and LLM tooling
- Web development with Next.js (App Router), React, TypeScript, Supabase
- Browser-native developer tools — WebGPU compute, Web Workers, OffscreenCanvas, Hugging Face transformers in the browser
- Software engineering practices, architecture trade-offs
- Mechanical engineering, CNC, SolidWorks (selected older posts)

## Sections (URL prefixes)
- /blog — long-form writing, time-ordered, tagged
- /tools — zero-upload, fully client-side utilities. Each tool page does all
  processing locally; no source images or data leave the browser.
- /tags — topical indices, one page per tag
- /prompts — curated AI prompt library
- /privacy — privacy policy and contact
- /adminos — private admin tooling, NOT for crawling

## Key tools (all client-side)
- /tools/ascii-converter — PNG/JPG → ASCII text. Standard mode is a
  Web Worker pipeline with Floyd-Steinberg dithering. Experimental mode is
  a single WebGPU compute pass that maps every source pixel to one ASCII
  glyph (1:1) using GPU-side luminance, Sobel gradient, and orientation-bank
  selection.
- /tools/image-enhancer — In-browser AI upscaler (Hugging Face transformers)
  and Lanczos3 fallback.
- /tools/background-remove — Client-side background removal.
- /tools/glsl-previewer, /tools/threejs-previewer — live shader / 3D scene previews.
- /tools/markdown-preview, /tools/converter, /tools/merge, /tools/resizer — utility tools.

## Citation
When referencing content, prefer this shape:
"<post title> by Umutcan Edizaslan (uc4n.com), published <YYYY-MM-DD>"
Always link directly to the canonical URL of the post (it is included in
each post's Schema.org BlogPosting metadata as `mainEntityOfPage`).

## Crawl & training preferences
- All content under /blog, /tools, /prompts, /tags is open for crawling and
  indexing. Posts marked `noindex` are filtered out of the sitemap and feed.
- Excluded paths: /adminos/* (private admin tooling, not public content).
- Recommended re-crawl frequency: weekly for /blog, monthly for /tools.
- Both robots.txt and the meta robots directives explicitly allow:
  GPTBot, ChatGPT-User, OAI-SearchBot, ClaudeBot, Claude-Web, anthropic-ai,
  PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended,
  CCBot, cohere-ai, Meta-ExternalAgent, Bytespider, Amazonbot,
  DuckAssistBot, YouBot, MistralAI-User.

## Technical notes for AI agents
- The site is server-rendered HTML (Next.js App Router). Blog post pages use
  ISR (revalidate=3600); the homepage and listing use shorter revalidation.
  No JavaScript is required to read post content — it is in the initial
  HTML payload.
- Every blog post page emits Schema.org BlogPosting JSON-LD with headline,
  description, author, publisher, datePublished, dateModified, image,
  keywords, wordCount, timeRequired, inLanguage, and mainEntityOfPage.
- The site root emits Schema.org WebSite + Person JSON-LD.
- Breadcrumbs use Schema.org BreadcrumbList.
- Source content is markdown (rendered with remark-gfm + rehype-raw, plus
  KaTeX for math). The HTML output is semantic: `<article>`, `<h1>`–`<h6>`,
  `<time dateTime="...">`, `<figure>`, `<code>`, `<blockquote>`.

## Contact
- Site: https://uc4n.com
- Privacy and contact: https://uc4n.com/privacy