Block rm -rf with a Claude Code PreToolUse hook, and know its limits

<p><em>This article was originally published on <a href="https://etherealogic.ai/exit-code-2-how-claude-hooks-turn-agentic-rules-into-runtime-barriers/" rel="noopener

<h1> Your Claude Plugin Marketplace Needs More Than a Git Repo </h1> <p>Anthropic's <a href="https://github.com/anthropics/claude-plugins-official" rel="noopener

<p>Claude Code is powerful. <br /> It can also silently write to your .env or run rm -rf.<br /> You find out after it happens.</p> <p>Waymark is an MCP server that intercepts<br

<p><code>git_commit.py</code> in my project is a small script: it reads the staged diff, sends it to <code>claude -p</code> with a system prompt asking for a Conventional Commit

<p><em>In retrospect, it was pretty obvious... sharing it anyway.</em></p> <p>I'm currently working on a self-steering progressive disclosure system, and I wanted my agent to read

<p>Claude Code's Stop hook blocks the agent from finishing until verification passes. Combine it with PostToolUse feedback injection to build a 3-layer verification loop (syntax,