State of Coding — Week of Aug 09: One Release Just 32x'd What AI Can Read

📄 Full written solution: https://interview-kit-fe.vercel.app/state-of-coding-2026-08-09 💻 Code on GitHub: https://github.com/timepasshub2539-hue/leetcode-python/tree/main/state-of-coding-2026-08-09 Chapters: 0:00 One Release Just 32x'd What AI Can Read 0:07 Kimi K3 Ships 256K Context 0:18 Old Prompt vs New Prompt 0:46 The Number That Matters 0:57 Watching Adoption Climb 1:06 tuicr Speeds Up Terminal UIs 1:17 One Year Fighting Scrapers On 1.5 Million Pages 1:28 Big Context Isn't Free 1:39 Fundamentals Still Win 1:51 This Week In One Line This week: one release just multiplied how much context AI models can read by 32x, and Kimi K3 shipped a 256K context window right behind it. We compare an old prompt against a new one to see what actually changed, dig into the one number that matters more than the headlines, and watch adoption climb in real time. Plus: tuicr speeds up terminal UIs, one year and 1.5 million pages of fighting scrapers, why big context isn't free, and why fundamentals still win. This week in coding, in one line. #coding #ai #llm #softwareengineering #techweekly Watch next: - Why Hard-Coding Numbers Ruins Your Reports #shorts: https://youtu.be/eb9SZqlICqI - Python Broke My print() Statement (On Purpose) #shorts: https://youtu.be/rnjOftVb84s - 621 vs 612: Decoding the Clue That Confused Me: https://youtu.be/WihI3tbhym0

# before: manual chunking
for chunk in split_repo(size=8000):
    ask(chunk, "where's the bug?")

# after: just paste it
ask(full_repo, "where's the bug?")

Keep going

0:45Free

deepseek-harness — trending on GitHub

General

6:31Free

Longest Subsequence With Non-Zero Bitwise XOR

General