Juejin
24 articles tagged with this topic
AI Debugging Is Smoother — But Devs Are Handing Over Their Secrets
Devs paste DB passwords, user data, and prod logs into AI for debugging—a Juejin breakdown lists 3 sensitive data types and 4 pre-commit checks.
Chinese Dev Runs AI Test Pipeline — Orchestration Is What Actually Changes Work
Chinese developer chains four AI test skills into one automated pipeline. Single AI capabilities save time; orchestration changes how work actually ge
AI Agent Crashes on Errors? Feed Failures Back to the Model, Don't Harden Code
'Error as Observation' feeds 403s, timeouts, and tool failures back to the LLM, with dynamic circuit breakers—explains why enterprise Agent demos brea
AI Engineers Now Have Their Own Toolkit: 44-Piece List Exposes a New Profession's Workflow
A 44-tool Mac setup list, from Cursor to Claude Code to Cloudflare—what it's really saying: AI development is shifting from geek experiment to a stand
TRAE Work Cuts Publishing from 2 Hours to 20 Minutes—But the Pipeline Won't Replicate
A writer used ByteDance TRAE Work to cut multi-platform publishing from 2 hours to 20 minutes via a 3-role AI pipeline. Most can't replicate it.
Senior Dev Built an App With AI End-to-End—The Real Deliverable Is the Workflow
Senior dev used AI end-to-end on a GitHub client. The app isn't the point—the deliverable is an 'AI dev scaffold' for product, UI, architecture.
Juejin Launches 'Vibe Works Plaza' — AI Coding Finally Has Its Own Showcase
Juejin, a top Chinese dev community, launched an AI-coding showcase with a ~10K RMB prize. Signal: AI coding has become a standalone track.
'No Errors' ≠ 'No Bugs': Five-Layer AI Code Acceptance Is the New Baseline
A Juejin AI-coding tutorial hit 10K reads—top question wasn't 'how' but 'why still buggy.' Skipped acceptance. We unpack the five-layer framework.
AI Code Generation 10x Speedup Is Real — But Only After You Lock Down the Rules
Framework author hit 10x efficiency with LLM code gen — but admits templates, not AI, do the real work. A must-read for any AI efficiency team.
AI Coding Answers Wrong? The Problem Is Your Prompt: 4 Templates
More people use AI to write code, but more get off-target answers. The problem isn't the AI — it's your prompt. Four beginner-ready templates inside.
AI Long-Form Writing Gets Formulaic — A Developer's 'Planned Derailment' Fix
A developer fixed AI long-form writing's self-copying habit at chapter 20 with 8 triggers + 6 disruption actions. Implications stretch beyond web fict
Dev Builds Mini Agent Framework in 8 Plugins — AI Infrastructure Moat Is Leaking
Harness reconstructed in 8 plugins, under 200 lines. When one dev builds an Agent skeleton in days, every Agent platform must recalculate moat depth.
After a year of AI coding, senior developers admit: judgment beats prompts
Juejin's AI coding series sums up 6 days in 3 principles: AI isn't an answer machine, write clear specs, verify all output. Real barrier isn't tools—i
AI Solved Writing. Auto-Publishing to Three Platforms Is the Real Bottleneck
AI drafts auto-publish from Obsidian to three platforms. Bottleneck: visible failures, no system-wide stability—solo creators becoming micro-ops.
AI Coding's Real Bottleneck Is Workflow, Not AI: 5 Pitfalls Exposed
A Juejin article distills 5 AI coding pitfalls exposing deeper workflow flaws—not skill gaps, but process gaps in enterprise AI adoption.
Juejin Puts 'Escape AI Burnout?' on Top — 3 Signals More Real Than Models
Juejin's dev feed put 'Escape AI Burnout?' beside games, food, fashion, travel. AI anxiety and ordinary life share one leaderboard.
AI Discounted a ¥99 Item by 10% — Verification Is the New Bottleneck
A Juejin article argues runnable AI code isn't deliverable code. AI cut "writing" costs but pushed up "verification" costs — and that's where develope
AI Auto-Coding Keeps Failing — Not the Model. Your Loop Engineering Is Broken
Project success hinges on workflow, not model strength. Juejin breaks down three engineering cards via an Android login module. Practical for AI teams
AI Says 'Healthy' but Returns Empty Strings: The LLM Monitoring Blind Spot
Production incident: HTTP all 200, yet AI assistants quietly returned empty strings. Article details a three-layer LLM health check framework.
'Users Will Love It' Isn't a Product Goal — Juejin Post Exposes 3 AI Pitfalls
Juejin post argues products need 'hypothesis → validation → decision,' not 'users will love it.' Three named biases apply to today's AI launches.
Product architecture is the underrated bottleneck in the AI coding era
AI coding tools are fast, but product managers still fumble in review meetings. A viral Juejin article argues 1 table + 4 diagrams is what actually ma
Time Saved by AI Coding Is Eaten Up by Manual Review — A PHP Developer's Reinvention Diary
A PHP engineer using Claude Code and Codex to pivot to Golang shipped zero code in episode 58 of his series, after manually rewriting AI-generated con
Enterprise RAG in Practice: The Bottleneck Is Docs and Retrieval, Not Models
A real enterprise deployment shows RAG succeeds or fails less on model size than on document parsing, chunking, retrieval, and citations.
Systematic Debugging Guide: A Detective Framework for Root Cause Analysis
A Chinese developer tutorial outlines a four-phase systematic debugging methodology replacing ad-hoc fixes.