Skip to content

Instantly share code, notes, and snippets.

View yigitkonur's full-sized avatar
🗿
Work hard, play soft

Yigit Konur yigitkonur

🗿
Work hard, play soft
View GitHub Profile
@yigitkonur
yigitkonur / turkish-search-fold.md
Created November 6, 2026 21:53
Turkish search folding rule (searchFold, test table)

Turkish search folding (one rule for every MCP)

Context: CLI tests showed GMC brand search ignoring diacritics ('gural', 'armagan' found nothing) and case-sensitive filters, and Search Console matching 'istanbul' (4 hits) but not 'İstanbul' / 'İSTANBUL'. Our users are Turkish SEO consultants.

Spec

Every MCP that matches user-typed text against names, titles, domains, brands, keywords or properties MUST compare searchFold(query) against searchFold(candidate).

Fold steps, in this order:

  1. Map İ, I, ı to i BEFORE any case mapping (so İ, I, ı, i all become i).
@yigitkonur
yigitkonur / cloudflare-service-selection-guide-for-ai-agents.md
Created October 21, 2026 20:20
cloudflare service selection guide for ai coding agents: workers, d1, r2, containers, cost gates, opt-in security. so your agent stops containerizing everything

Cloudflare service selection guide for AI coding agents (Workers, D1, R2, Containers, cost & security)

hi human. this file is not really for you, it's for your agents. i got tired of watching coding agents spin up a container for a hello world, bolt a waf onto a todo app "just in case", and mix up workers ai, workers for platforms and workers vpc like they're the same thing. so i wrote down how i want them to choose cloudflare services. you can read it too, it's free.

tl;dr (for humans)

  • default stack is cloudflare workers + static assets, d1, kv, r2, durable objects, queues and workflows: cheap, pay-per-use, scales to zero.
  • containers, sandboxes, browser run, stream, images and ai search sit behind a cost gate. an agent has to say why the defaults can't do it, plus a rough usage estimate.
  • security is opt-in. turnstile, rate limiting, access and worker secrets are the cheap sensible ones; the rest only on explicit request or a real threat.
  • there's a "look-alikes" section so agents stop confu
@yigitkonur
yigitkonur / VibeScroll-Support-Privacy.md
Last active July 9, 2026 06:57
VibeScroll App Store support and privacy

VibeScroll Support and Privacy Policy

Last updated: July 9, 2026

Support

VibeScroll is a native macOS app for viewing short-form video feeds in live floating WKWebView windows with native Mac controls.

For support, email yigit35@gmail.com with "VibeScroll" in the subject, or use the support flow provided by the Mac App Store listing.

@yigitkonur
yigitkonur / linkedin-to-md.py
Last active December 8, 2025 22:10
Raycast script: LinkedIn Profile to Markdown via Clado API - fetches LinkedIn profiles, generates structured Markdown with experience, education, skills. Supports multiple URLs in parallel. Uses MD cache.
#!/usr/bin/env python3
# Required parameters:
# @raycast.schemaVersion 1
# @raycast.title LinkedIn Profile to Markdown
# @raycast.mode silent
# Optional parameters:
# @raycast.icon 💼
# @raycast.packageName AI Tools
@yigitkonur
yigitkonur / urls-to-md.py
Created December 8, 2025 21:42
Raycast script: Parse URLs with Jina AI - reads URLs from clipboard, processes in parallel, copies formatted markdown content
#!/usr/bin/env python3
# Required parameters:
# @raycast.schemaVersion 1
# @raycast.title Parse URLs with Jina AI
# @raycast.mode silent
# Optional parameters:
# @raycast.icon 🌐
# @raycast.packageName AI Tools
@yigitkonur
yigitkonur / youtube-transcribe.py
Last active December 8, 2025 21:45
Raycast script: YouTube Transcribe with ElevenLabs Scribe - downloads audio, transcribes with diarization, generates SRT subtitles and speaker-formatted output
#!/usr/bin/env python3
# Required parameters:
# @raycast.schemaVersion 1
# @raycast.title YouTube Transcribe with Diarization
# @raycast.mode silent
# Optional parameters:
# @raycast.icon 🎙️
# @raycast.packageName Media Tools
@yigitkonur
yigitkonur / clipboard-to-md.py
Created December 8, 2025 21:29
Raycast script: Clipboard HTML to Markdown converter - converts rich HTML from clipboard to clean Markdown
#!/usr/bin/python3
# Required parameters:
# @raycast.schemaVersion 1
# @raycast.title Clipboard HTML to Markdown
# @raycast.mode compact
# Optional parameters:
# @raycast.icon 📋
# @raycast.packageName Clipboard Tools
@yigitkonur
yigitkonur / duti.sh
Created November 13, 2025 00:01
duti.sh to open all files on sublime
#!/bin/bash
################################################################################
# Comprehensive Sublime Text 4 Default File Association Script
#
# This script sets Sublime Text 4 as the default application for a wide range
# of file extensions commonly used in software development, configuration,
# documentation, and data processing.
#
# Requirements: duti (install via: brew install duti)
@yigitkonur
yigitkonur / comprehensive-eslint-rules.md
Created August 13, 2025 07:01
a context for guiding LLM to choose best ESLint rules for your projects

A Strategic Framework for Code Quality: From Linting Rules to Production Resilience

Let's try to understand each of them by reviewing that summary:

ESLint Core "Possible Problems" Rules (Summary Table)

Rule Name Core Problem Prevented Fixable (🔧) Recommended (✅) Primary Production Benefit
array-callback-return Unexpected undefined values from array methods. 🔧 ✅ Data Integrity
constructor-super ReferenceError in derived class constructors. ✅ Stability
@yigitkonur
yigitkonur / llm-mermaid.md
Created June 3, 2025 04:57
A prompt for forcing AI to use mermaid more effectively

Your primary function is to transform ANY textual diagram idea, natural language description, malformed/incomplete Mermaid code, or embedded Mermaid blocks within Markdown into production-ready, syntactically pristine, visually compelling, and interactive Mermaid diagrams. You will also provide micro-documentation via a concise changelog and embedded tooltips. Your core operational logic is derived from the comprehensive Mermaid syntax and feature compendium detailed herein.


I. OPERATIONAL PHASES (Your Refinement Lifecycle)

Phase 1: Input Ingestion & Contextual Analysis

  1. Isolate Mermaid Content: If input is Markdown, extract content from ```mermaid ... ``` blocks. For other inputs, identify the core diagram-related text.
  2. Pre-sanitize: Normalize basic whitespace; identify explicit user flags (theme:, type:, layout:).
  3. Diagram Type & Layout Inference (See Section II: Inference Matrix): Determine the most appropriate Mermaid diagram type and initial