Posts
All the articles I've posted.
-
Kimi K2.7 Code Lands in GitHub Copilot: After Benchmarks and Buzz, a Chinese Model Finally Wins the Channel
GitHub Copilot added Kimi K2.7 Code to its model picker on 2026-07-01 — the first open-weight and first Chinese model there. Hand-verified data, plus the "open weights + Azure hosting + off-by-default" trust structure.
-
WorkBuddy Hit #1 in China in 90 Days: What messaging-first AI Agents Teach the Claude Code / Codex CLI World
Updated:Tencent WorkBuddy launched 2026-03-09 and hit 8.85M monthly visits (+831% MoM) with a DAU 3-4x the
-
GLM 5.2 Beats Claude on a Cyber Benchmark: Is China's 'Scenario-Specific Superiority' the New Inflection Point?
Updated:Zhipu's GLM 5.2 hit 39% F1 on Semgrep's IDOR vulnerability detection benchmark, beating every Claude Code Opus version at roughly 1/6 the per-finding cost — while still trailing about 2 points on aggregate benchmarks. Kimi and Qwen reproduce the same pattern. Is "scenario-specific superiority" the 2026 inflection point for Chinese models?
-
AI Model Comparison Mid-2026 Sequel: 52 Days After blog166, Here's What I Got Wrong in May
A line-by-line self-audit 52 days after blog166: of 5 predictions, 2 retracted (the Sonnet/Opus advice is stale, the open-source gap was underestimated), 1 still pending (Anthropic pricing), 2 held up. Plus six months of real model-routing observations from this blog and the checkpoints for the next audit.
-
Ponytail Hit 73k Stars in Three Weeks: A Side Project Cured AI Agents' Over-Coding Disease
DietrichGebert's ponytail launched on 2026-06-12 and climbed to 73k stars / 3,848 forks in three weeks, covering 16 agent ecosystems. Its core is a 7-step decision ladder that forces AI to obey YAGNI. I ran an A/B test against a 196-line TableOfContents component from this blog; after installing ponytail the review recommendation was to cut it to 25-30 lines, a -84% change. Inside — firsthand test data, three structural reasons for the viral growth, a step-by-step breakdown of the ladder, and three counter-boundaries.
-
WeRead's Official Agent Skill: The First Chinese Content Platform to Ship One — But Is the Indie Dev Monetization Window Actually Open?
微信读书 (WeRead) shipped its official Agent Skill on 2026-05-17 — the first "official player" among Chinese content platforms. Community forks exploded in the first week; six weeks later there is still not a single commercialized product. This post unpacks the three structural blockers behind the "official access + community boom + zero commercialization" pattern, and what indie devs can actually cash in on right now.
-
ZCode on HN's Front Page: When Silicon Valley Actually Read a Chinese Dev Tool's Docs
Zhipu's official GLM-5.2 harness "ZCode" hit the Hacker News front page (155/192 when I first opened it, 264/236 now). This isn't "yet another AI IDE" — it's the first time a Chinese dev tool has been seriously evaluated by mainstream Silicon Valley engineers. This post breaks down the timeline, what ZCode actually is (a desktop app, not a CLI), the three real objections on HN, and a 30-line TS snippet for wiring up GLM Coding API yourself.
-
The 90-Day Spaghetti Point: Why Maintenance Is Vibe Coding's Real Exam
When Karpathy coined Vibe Coding in Feb 2025, he only talked about how good it felt to write. GitClear's 211M-line study and the 2026 wave of rescue jobs both map that feeling onto a 4x maintenance bill in year two. This post unpacks the three mechanisms that detonate around day 90 — the Spaghetti Point — and how to keep AI speed without paying that bill.