Copy this rig quick start skills (5) guardrails only
# review before running: this installs third-party code
$ npx degit husterk/dotfiles/.claude ./rig-dotfiles # inspect, then merge into .claude/ copy MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ npx degit husterk/dotfiles/.claude/skills/add-app .claude/skills/add-app
$ npx degit husterk/dotfiles/.claude/skills/remove-app .claude/skills/remove-app
$ npx degit husterk/dotfiles/.claude/skills/start-work .claude/skills/start-work
$ npx degit <redacted>-repo .claude/skills/audit-repo
$ npx degit <redacted> .claude/skills/unslop copy Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Edit(hosts/*/generated/**)"
]
}
} copy json Skills (5) Permissions deny (1)
Edit(hosts/*/generated/**)
allow (5)
Bash(mise run check)
Bash(gh issue list *)
Bash(gh issue view *)
Bash(gh issue create *)
Bash(gh issue comment *)
Context tax
Instructions: 1.0k MCP tool schemas (est.): 0 Skill metadata: 423 Subagent metadata: 0
ALWAYS-LOADED FILES (claude-code)
Estimate: characters ÷ 4 for instruction files and skill/subagent descriptions; MCP schema sizes from typical tool counts. Harnesses with deferred tool loading pay less. Method
Guardrails 1/5 · quality 15/100 ✗ No deny/ask rule for rm -rf, force-push, reset --hard, sudo…
✗ No deny rule for .env / keys / ~/.ssh
✓ Permission prompts stay on
✗ No sandbox and no ask-first rules
Destructive-command coverage 0/22 dangerous variants covered by deny/ask rules or a blocking hook. Missed: rm -rf rm -r -f rm -fr rm -Rf find -delete git push --force git push -f git push --force-with-lease git reset --hard git clean -fdx sudo chmod -R 777 dd mkfs curl | sh terraform destroy kubectl delete npm publish DROP TABLE Known bypasses not caught: /bin/rm -rf (absolute path) bash -c 'rm -rf …' git -C . push --force
×35% 0
Secret protection breadth Not denied: .env* (Read), .env* (Edit), *.pem / *.key (Read), *.pem / *.key (Edit), ~/.ssh (Read), ~/.ssh (Edit), ~/.aws (Read), ~/.aws (Edit)…
×25% 0
Hook effectiveness no PreToolUse hook
×15% 0
No-YOLO + allow-list hygiene No YOLO mode, no broad allows
×15% 100
Instruction clarity Explicit never/always rules found for: none · missing: deletion, secrets, git, prod
×10% 0
Deterministic static analysis, identical in the crawler and here. Nothing is executed. Hooks whose scripts live outside the repo are "unknown", not a pass.
Add the badge Show this rig's score in your README. Paste this Markdown:
[](https://agentrigs.dev/r/husterk/dotfiles/?ref=badge) copy The badge updates by itself after each daily crawl. What the score means · remove this rig
TO RAISE THIS SCORE
Add deny rules for destructive commands to .claude/settings.json, e.g. "Bash(rm -rf:*)", "Bash(git push --force:*)", "Bash(git reset --hard:*)". Deny reads of secrets, e.g. "Read(./.env*)", "Read(~/.ssh/**)", "Read(**/*.pem)". Add a PreToolUse hook that checks Bash commands and exits 2 to block the dangerous ones. Indexed from commit 02f915f4515d on 2026-11-17. Data is structure only — env values and anything secret-looking are stripped.