Forty-five ways to stop babysitting Claude Code

What it does This repo collects 45 concrete tips for squeezing more autonomy out of Claude Code, Anthropic’s terminal-based coding agent. It covers everything from voice transcription and context compression to running the agent inside containers and wiring Gemini CLI as a fallback for blocked sites. The repo also ships a custom context-bar.sh status-line script and a dx plugin for everyday developer workflows.

The interesting bit The advice reads like a battle log from someone who actually lives in the tool—there are tips for whispering voice commands on airplanes, manually implementing exponential backoff for long Docker builds, and using Git worktrees to keep parallel Claude sessions from stepping on each other. It treats Claude Code less like a search engine and more like a contractor who needs clear deliverables and tight feedback loops.

Key highlights

• Custom status line script shows model name, git state, token-usage progress bar, and your last prompt so you don’t lose the thread between sessions
• Voice workflow guide covers local transcription tools and how to whisper commands quietly enough for open offices (and airplanes)
• Includes a dx plugin with skills for routine dev workflows, plus a quick-setup script
• Explains how to slim the system prompt, run subagents in containers, and use Gemini CLI as a “minion” when sites are blocked
• Covers practical slash commands like /usage, /stats, and /mcp with real output examples

Verdict Grab this if you’re already using Claude Code daily and want to graduate from basic prompts to a personalized, automated workflow. Skip it if you’re looking for a packaged CLI tool—this is a field manual, not a framework.