repo-armor free — 3 Claude Code hooks (dangerous-command blocker + main-branch guard + notifications)
Deterministic guard rails for Claude Code — hooks are shell scripts that run on every tool call, outside the model, so they hold even when the model has a bad day. This free pack contains 3 of the 10 repo-armor hooks:What you get (free): block-dangerous-commands.sh — hard-blocks rm -rf on paths that matter, force-pushes to main, git reset --hard, DROP TABLE, curl|sh — with the reason fed back so the agent picks a safe path guard-main-branch.sh — blocks commits/pushes on main/master (override with ALLOW_MAIN=1) notify.sh — desktop notification (macOS/Linux) when the agent needs you Design: plain bash + jq you can read in minutes. Fail-open — if jq is missing or a script errors, your session never breaks. No network. No telemetry. Env-var kill switch per hook.Want the full armor? The complete pack (10 hooks) adds: secret-leak blocker (AWS/Anthropic/OpenAI/GitHub/Slack/PEM), protected-file guard (.env, lockfiles, migrations), test-on-stop (the agent can't declare "done" with red tests), format-on-edit, lint feedback loop, JSONL audit log, context guard — plus a one-command installer with settings merge/backup/uninstall and per-stack configs. 52 self-tests, shellcheck-clean.→ Get repo-armor (all 10 hooks, $5): https://solemness.gumroad.com/l/repo-armor
Get it → solemness.gumroad.com