Evaluator-Optimizer in Claude Code: From Pattern to Skill
How I turned an Agentic AI pattern into a reusable Claude Code skill with evidence-based evaluation.
Claude Code
From zero to pro with Claude Code. In-depth guides and tips you can apply today.
8 articles · 33 tips
Articles
Claude Code Hooks: Practical Guide with Real Examples
Practical guide to Claude Code hooks: 3 hook types, 5 essential hooks, advanced patterns and common pitfalls. With copy-paste ready examples.
Claude Code Skills: Build Reusable Workflows with Custom Commands
Skills are the fastest way to create repeatable workflows in Claude Code. A markdown file, a slash command, and predictable steps with human intervention.
Automated Code Review with Claude Code, Playwright, and Notion (MCP)
Learn how to connect Claude Code with Playwright and Notion via MCP to create an automated UI and code review system.
Claude Code Subagents: Practical Guide with Real Agent Configs
Learn to create and manage subagents in Claude Code. Divide complex tasks, avoid context contamination, and scale your AI development.
Claude Code Guide: Professional Setup, Configuration & Workflows (2026)
The definitive Claude Code guide (2026). Installation, CLAUDE.md configuration, hooks, skills, subagents, and professional workflows for developers.
MCP in Claude Code: The Practical Setup Guide (2026)
Understand what the Model Context Protocol (MCP) is and how to use it in Claude Code to connect your LLM with real tools (GitHub, Linear, Databases).
Quick Tips
Stop Wasting Tokens: Pick the Right Model in Claude Code
Use opusplan in Claude Code: Opus for planning, Sonnet for execution, Haiku for exploring. Stop using the frontier model for everything.
When Does Claude Code Load Each Feature (and What It Costs)
Learn when each Claude Code feature loads into context — CLAUDE.md, Skills, MCP, Subagents, and Hooks — and how to save tokens from the start.
Claude Code Desktop: 5 Features Worth Leaving the Terminal For
The 5 exclusive features of Claude Code Desktop over the CLI: visual diffs, parallel sessions, embedded preview, PR monitoring, and built-in connectors.
The Agentic AI Pattern That Always Keeps You Honest
Build the Evaluator-Optimizer pattern as an inline skill in Claude Code. One LLM generates, another evaluates every claim against real evidence.
4 Agentic AI Patterns You Already Use in Claude Code — and One You Don't
Every agentic pattern from Anthropic's guide maps to a Claude Code primitive. Here's the complete reference.
Claude Code Hooks — Automate Your Workflow
Hooks are commands that run automatically at specific points in Claude Code's lifecycle. Unlike CLAUDE.md instructions, hooks are deterministic: if the condition is met, they always execute.
Quick Tip: Automate Skill Permissions with allowed-tools
Use the allowed-tools field in your Claude Code skills to automate tool approval and eliminate repetitive confirmation prompts.
Put 3 Claudes to Work at Once (One Command)
claude -w feature-auth creates a git worktree, branches your code, and starts a new Claude instance in it. Run it three times in three terminals and you have three Claudes working on three different tasks — in parallel, on the same repo, without conflicts.
Your Claude Code Session Has Branches (And You Didn't Know)
claude --continue --fork-session creates an independent branch of your current conversation. Same context, new path. Like git branch, but for your sessions with Claude.
Claude Code Opens Chrome and Debugs Your Frontend for You
claude --chrome or /chrome connects Claude Code to your browser. Claude opens tabs, navigates, clicks, reads the console, and debugs your frontend — all without leaving the terminal.
Show Thinking in Claude Code: Verbose & Extended Mode
Press Ctrl+O to watch Claude think. Spot problems early. Stop bad decisions before they become bad code.
What Is MCP in Claude Code and How to Set It Up in 2 Minutes
MCP is not a plugin, not a skill, not an API. It's a protocol. claude mcp add + URL or local command and Claude is connected to any service.