claude

guardrails

Guardrail hooks, a plan → implement → verify → crosscheck loop, a /critical-review exit gate with a pluggable reviewer backend, doc-tiering commands, a /setup-check, a browser-MCP skill and a terse output style for Claude Code.

31 on GitHubby Till Meierv0.5.3MITupdated 7 days agoproductivity
/

Plugin guardrails from marketplace tillmeier in tillmeier/claude-code-guardrails. Commands read from the repo's manifest — 6 days ago.

This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.

What it installs

Skills (1)

  • browser-mcp

    Read before the first chrome-devtools-mcp call in a session (an isolated server such as mcp__chrome-devtools__*, or one attached to your real Chrome). Which server for what, token-cheap check order, filePath roots, wait_for / emulate / screenshot pitfalls, tab hygiene, secrets in pages, browser memo

Slash commands (10)

  • /guardrails:cleanup-docs

    Audit and clean up a repo's Claude documentation — measure the always-on tier first (CLAUDE.md + nested + .claude/rules/ + the auto-memory index), then merge redundancy, cut derivable and model-obsolete content, demote misplaced tiers, fix cross-refs. Verifies every fact survives before cutting; ask

  • /guardrails:critical-review

    Exit-gate review of this session's changes by a reviewer that never sees the task brief — Codex, Claude or your own command via scripts/review-backend.sh. Findings → plan mode with proposed fixes (one approve/deny), or --review-only for findings in chat. --adversarial forces the deeper prompt-mode p

  • /guardrails:crosscheck

    Cross-check recent, not-yet-verified work — in ANY project. Reconstruct what you shipped or optimized lately and haven't confirmed, find where each change's evidence lives (logs / DB / generated output / endpoint / UI / CI — discover, don't assume), and report honestly: does it run clean, did new pr

  • /guardrails:implement

    Execute an already-agreed plan one step at a time — implement the smallest shippable increment, verify it before moving on, ask open questions instead of guessing, and HALT on anything unforeseen. Pass a plan file as the argument, or omit it to use the plan already in the conversation.

  • /guardrails:init-local-md

    Create or reinitialize .claude/local.md with environment-specific configuration. Detects project type, environment, and available tools, then writes an appropriate template. Pass `--force` to overwrite without backup prompt.

  • /guardrails:plan

    Front-load a task with a short, cited design spec and get approval BEFORE writing code. Produces a persistent plan artifact that survives compaction and defines the verification contract /verify later checks. Pass the task description as the argument; add --codex to have Codex adversarially review t

  • /guardrails:refactor-claude-md

    Refactor a project's Claude documentation to the distributed pattern — lean CLAUDE.md core + on-demand tiers (.claude/docs/ feature details, .claude/rules/ path-scoped conventions, nested per-subsystem CLAUDE.md, .claude/local.md environment), with an auto-generated doc index and no @-imports. Detec

  • /guardrails:save-learnings-to-docs

    Save session learnings to the CHEAPEST tier that carries each one (hook → .claude/rules/ path-scoped → .claude/docs/ → skill → .claude/local.md → auto memory; CLAUDE.md last and gated), enforcing the CLAUDE.md and MEMORY.md budget gates, AND reconcile the docs related to this session — remove redund

  • /guardrails:setup-check

    Check what of guardrails is set up on this machine — git/python3, the /critical-review reviewer, the Codex plugin, Node, Chrome, browser MCP servers, output style, hooks registered twice — and print the exact fix for each gap.

  • /guardrails:verify

    Ship gate. Scope to this session's diff, exercise the change end-to-end (drive the real flow, not just tests/typecheck), and emit a PASS / FAIL / SKIP verdict with evidence. Run before /save-and-push on anything with a runtime surface. Optional focus hint as argument.

Hooks (6)

  • SessionStart

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh"

  • SessionStart

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/compact-handoff-inject.sh"

  • PreToolUse

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/sql-guard.sh"

  • PreToolUse

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/git-staging-guard.sh"

  • PostToolUse

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/format-file.sh"

  • PreCompact

    bash "${CLAUDE_PLUGIN_ROOT}/hooks/precompact-handoff.sh"

View source on GitHub

Similar plugins

superpowers

claudecodex

Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques

15 skills · hooks

obra/superpowers 316,836

caveman

claude

Talk like caveman. Cut filler. Keep technical details explicit.

22 skills · 1 command · 3 agents · hooks

JuliusBrussee/caveman 119,462

claude-mem

claudecodex

Persistent memory for Claude Code: the next session starts where the last one left off.

29 skills · 1 MCP server · hooks

thedotmack/claude-mem 116,618

Taste Skill

claudecodex

tasteskill 2.5.1: professional company websites at the figma.com and x.ai level, with one section per screen, real photographs, real product UI, effects kept for tech and gaming pages, and a real-browser checker and finish gate.

1 skill

Leonxlnx/taste-skill 114,391

career-ops

claudecodex

Open-source AI job search that runs locally in your AI coding CLI: scan job portals, evaluate listings into a structured A-H report with a 1-5 score, tailor your CV, track applications

5 skills · hooks

career-ops-hq/career-ops 82,253

mem0

claudecodex

Cross-session memory and token savings for coding agents.

6 skills · 1 agent · 1 MCP server · hooks

mem0ai/mem0 68,893

Maintain guardrails?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex pluginslisted