Alon Wolenitz

Alon Wolenitz

Developer sharing honest takes on AI-assisted development, sandboxing, and the tools that actually move the needle. Writing from the trenches of production codebases.

15 Posts

Posts by Alon Wolenitz

Quick Bash Wins: The ! Prefix

Stop wasting tokens asking Claude to run commands. The ! prefix executes bash instantly and injects output into context—no model processing, no delay, no wasted tokens.

Read more →

Claude Code Templates: 400+ Ready-to-Use Configurations

Claude Code Templates (aitmpl.com) provides 400+ agents, commands, templates, and integrations to accelerate your Claude Code projects. Browse, select, and install in minutes.

Read more →

tweakcc: Customize Everything in Claude Code

tweakcc is a CLI tool that brings customization to Claude Code—from system prompts and themes to toolsets and session naming. Here's what it does and how to get started.

Read more →

Enhanced Plan Mode in Claude Code: What Changed and Why It Matters

Claude Code v2.0.51+ introduced multi-agent plan mode with Haiku exploration. Here's what changed, the problems the community discovered, and when it works well vs when it doesn't.

Read more →

Fixing Enhanced Plan Mode: Community Workarounds That Actually Work

Struggling with plan mode's token consumption or quality issues? Here are four community-tested workarounds: custom commands, hooks, downgrades, and when to use each.

Read more →

Faster @ File Suggestions: Claude Code v2.0.55's Fuzzy Matching Upgrade

Claude Code v2.0.55 improved fuzzy matching for @ file suggestions, making it faster and more forgiving when typing partial filenames with typos or missing characters.

Read more →

Advanced Tool Use in Claude API: Three New Features That Change A lot

Anthropic introduces Tool Search, Programmatic Tool Calling, and Tool Use Examples - achieving 85% context reduction and 37% token efficiency gains.

Read more →

New Tools in Claude Code 2.0.51

Claude Code 2.0.51 introduces EnterPlanMode, agent hooks, and Plan Mode v2 with Opus 4.5 integration for enhanced development workflows.

Read more →

Claude Code Can Now Explain Itself: The Claude Code Guide Agent

Claude Code v2.0.49 quietly added a claude-code-guide subagent that helps Claude understand how to use its own features. Here's what it does and how it works.

Read more →

Teleport Tasks to the Cloud with the & Prefix

Version 2.0.45 introduced a powerful feature: start any message with & in the Claude Code CLI to run it as a background task on Claude Code Web.

Read more →

Fighting 'AI Slop': How Claude Code Skills Transform Frontend Design

Anthropic's new frontend design skill fights generic AI aesthetics by teaching Claude to avoid purple gradients, Inter fonts, and cookie-cutter layouts - here's how it works and why it matters

Read more →

Claude Code is a Beast: Lessons from Rewriting 300k Lines of Code in 6 Months

A deep dive into the production-grade workflow that enabled one developer to solo rewrite 300k LOC with Claude Code - featuring skills auto-activation, dev docs systems, PM2 debugging, and the hooks that changed everything

Read more →

Prompt-Based Stop Hooks: LLM-Driven Control Flow in Claude Code 2.0.41

Claude Code 2.0.41 introduces prompt-based stop hooks - an intelligent alternative to bash scripts that uses Claude Haiku to make context-aware decisions about when to stop execution

Read more →

The End of 'Yes, Yes, Yes, Whatever': How Claude Code's Sandboxing Solves Approval Fatigue

A developer's honest take on why I was blindly clicking 'approve' 100 times a day until sandboxing saved my sanity (and my codebase)

Read more →

Output Styles: Claude Code's Most Underrated Feature (And Why The Community Fought to Save It)

In version 2.0.30, Anthropic deprecated output styles. Four days later, they brought it back. Here's why this simple feature is the key to transforming Claude Code from a coding assistant into 'Claude Anything'

Read more →