📰 Dev.to · jidonglab

An interviewer's breakdown of why strong engineers fail coding interviews, what the scorecard really measures, and how the debrief decides.

STT event timings lie about when you stopped talking. Why I measure voice-interview silence from the waveform with WebAudio instead.

ContextZip tracks cumulative token savings across sessions. See exactly how much AI context window space you've reclaimed over days and weeks.

Your AI coding agent has a fixed context window. Every line of CLI noise pushes out actual code context. Here's how to reclaim 60-90% of that wasted space.

I read 80,000 lines of Claude Code's TypeScript source to understand why the tool behaves the way it...

Most people open an unfamiliar repo and immediately ask Claude Code: "what does this codebase do?"...

ANSI escape codes are invisible to you in the terminal but they consume real tokens in your AI coding agent's context window. Here's how to strip them.

Django and Flask tracebacks are 90% framework internals that waste your AI's context window. ContextZip strips them down to your application code frames.

I benchmarked 102 common CLI commands and measured how much of their output is noise vs. signal for AI coding agents. The results were eye-opening.

I cloned the Codex repo and started reading. Not the README. Not the blog post. The actual Rust...

npm install dumps hundreds of deprecation warnings into your AI coding agent's context window. Here's how to filter them out automatically with one tool.

The full origin story of ContextZip: a fork of RTK that grew into a language-aware CLI output optimizer purpose-built for AI coding agents.

Yesterday morning I fixed a production bug from the subway. I typed a single message in Telegram, and...

I don't write Rust. I can read it well enough to catch obvious bugs, but I've never typed impl or fn...

RTK is one of the best tools in the Claude Code ecosystem. 28k GitHub stars, 60-90% token savings on...

Saju-to-Shorts video pipeline. Architecture prompt first, then scaffold, then review. 4 commits, zero manual coding.

Saju-to-video pipeline. 4 commits. Architecture → scaffold → review → ship. The prompting pattern that kept 5,800 lines coherent.

Multi-stack AI coding breaks down at scale. Here are the prompt patterns that kept Claude Code coherent across 4 commits and 2 languages.