About adhibuchori/agent-config-kit
Claude Code plugins that stop an AI agent from doing the things you would regret: force-pushing to main, deleting src/, reading your .env into the chat, or writing to the production database. The guards are hooks that really refuse (exit code 2), not advice in a prompt. Around them the kit installs the rules, check scripts and pull-request CI your stack needs, so the agent works the way your team does: plan, review, commit, open a PR, merge.