Make infra/CI/deploy changes that are reproducible, reversible, and least-privilege.
Use to safely clean up stale LOCAL feature branches left behind by the TDD harness once their slices have shipped — both merged branches and abandoned (closed-PR) branches.
Executes background tasks safely by dropping privileges and enforcing timeouts. Includes ISNAD signed manifest.
Safely append CSS styles from one file to another by identifying missing rules, adapting CSS variables, and verifying changes
Ensure AI agents work in an isolated Git worktree to prevent changes to the main working directory. Use when AI is about to make its first code modification in a session, or when…
Safe command execution for OpenClaw Agents with automatic danger pattern detection, risk assessment, user approval workflow, and audit logging.
Safe command execution for OpenClaw Agents with automatic danger pattern detection, risk assessment, user approval workflow, and audit logging.
Read and view files without permission prompts. Use bat for syntax-highlighted code viewing, eza for directory listings with git status, cat/head/tail for plain text.
Prevents destructive Git operations by requiring backup and confirmation. Blocks git reset --hard, git push --force on main, git clean -fd.
带安全策略的 OpenClaw Skill 安装器。安装前自动经过 ClawShield 扫描，支持策略控制（来源限制、模式阻断、风险分级审批），提供快照和回滚能力。
Für SAFE/KISS deutsche Adaption: ordnet Norm, Beweislast und Gegenargument; Ergebnis: Schnittstellenkarte mit Zuständigkeits- und Nachweisfragen.
Run the safe Drizzle migration ritual for a repo with a fragile migration directory: pick the next free prefix, generate, grep the SQL for destructive operations, enforce IF…
Prevent destructive operations using Claude Code hooks. Three modes — cautious (warn on dangerous commands), lockdown (restrict edits to one directory), and clear (remove…
Check whether an online store or seller is legit before you pay — and pay in a way you can get your money back if it isn't.
Master skill that runs the full PlanetScale safe best-practices assessment — inventory, engine review, Insights, Traffic Control, webhooks, schema recommendations, codebase…
根目录改名/移动后，安全批量更新全树绝对路径引用。只替换「路径形式」的旧根前缀，保留品牌语义名、GitHub URL、历史旧根引用；规避备份目录递归陷阱；用 robocopy /MIR 强制清空长路径深层残留。适用于任何「D:\OldRoot → D:\NewRoot」或「嵌套到新父目录」的目录重构。
Safely analyze and reorganize project structure with multi-stage validation, dry-run previews, and explicit user confirmation.
Safe Git pull with rebase workflow — check remote changes, stash, rebase, restore, handle conflicts, and push. Covers main/develop sync and feature branch rebase.
Format with Complete Confidence — Smart backup reminders, risk assessment, and post-format validation. Your data stays protected, every time.
Check for sensitive data in repository. Use when user asks to "check for sensitive data", "/safe-repo", or wants to verify no company/credential data is in the repository.
YC SAFE Agreement review and advisory skill for startup founders and lawyers. Use when user (1) uploads a SAFE agreement for review/comparison, (2) asks questions about how SAFEs…
Use safe-rm script instead of rm -rf. Prevents accidental deletion of protected system, config, and project paths.
Use when reverting phase or implementation work requires dependency-aware rollback and recovery planning
Use when changing code or data files with a script or shell command instead of the Edit tool - bulk replacements across several files, multi-line edits, inserting blocks, or any…
Fallback workflow for code execution when automated tools fail, using shell heredoc with verification
Use when designing any job, webhook handler, command, startup step, or loop that changes state. Assume it will run twice and that the last run may have died halfway.
Ensure Git worktree slugs can never escape the managed workspace or inject traversal.
Reference — rules for safely executing write/destructive operations against ad accounts. Always loaded into the performance-marketer agent.
Deep-dive AI Safety, NIST AI RMF, and algorithmic bias compliance engine. (v5.0.0)
ASEAN data protection compliance engine — VN, SG, TH, MY, ID, PH regulatory frameworks. (v5.0.0)
Deep-dive compliance engine for products targeting or affecting children (COPPA, FERPA, AADC). (v5.0.0)
Deep-dive GDPR & EU AI Act compliance engine for European market products. (v5.0.0)
Healthcare compliance engine — HIPAA, HITECH, FDA SaMD for HealthTech products. (v5.0.0)
Deep-dive US state-level privacy (CCPA/CPRA, VCDPA, CPA, TDPSA) compliance engine. (v5.0.0)
SafeBase integration. Manage data, records, and automate workflows. Use when the user wants to interact with SafeBase data.
SafeGraph integration. Manage data, records, and automate workflows. Use when the user wants to interact with SafeGraph data.
Use when AI-generated or AI-modified changes need a Code Health gate before commit, handoff, or pull request.
Safepay integration. Manage data, records, and automate workflows. Use when the user wants to interact with Safepay data.
Anthropic's method for training harmless AI through self-improvement. Two-phase approach - supervised learning with self-critique/revision, then RLAIF (RL from AI Feedbac — from…
Meta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal pla — from…
NVIDIA's runtime safety framework for LLM applications. Features jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxi — from…
Use when a lawyer user needs guidance on the professional-responsibility duty of competence as it applies to AI tool use in legal practice.
Use when assessing whether AI-generated legal output may constitute unauthorized practice of law (UPL) in a given jurisdiction, or when configuring the appropriate output mode…
Verify before you trust — model pinning, fallbacks, and runtime safety validation
Audit workplace safety compliance systems for OSHA regulatory gap analysis, inspection readiness assessment, permit-to-work management, lockout/tagout program compliance, confined…
NASA Power of 10 rules adapted for writing robust, verifiable code with context-appropriate rigor
Safely delete files or directories in a git working tree by routing deletions through `git rm` or `git clean` instead of `rm`, `rmdir`, `shred`, `unlink`, `find -delete`, or…
Route potential professional safety concerns through approved evidence and qualified-human escalation.
Prüft Folgen von Safety-Gate-Warnung, öffentlicher Produktwarnung, Reputationsrisiko und Korrekturkommunikation.
Неприкасаемые safety-инварианты для изменений, затрагивающих безопасное состояние, `TIM1`, `BKIN/BKIN2`, `STO`, вторичный путь запрета, watchdog, машину состояний и валидность…
Provides an out-of-band stop control (throttle/pause/stop) for autonomous or long-running agent loops via a .claude/KILL file the agent cannot disable.
Detect risk or uncertainty and pause execution. Use before or after plans, commands, diffs, or implementation to surface unclear requirements or risky actions.
Use when a user appears to be under 18 or when a legal matter involves a child as a party, witness, or subject.
Use to govern when, where, and how the legal-information-not-legal-advice disclaimer must appear. Defines the precise line between permissible legal information and regulated…
Bounds blast radius when a session ingests untrusted input (web pages, GitHub issues, scraped content, MCP output) via the Agents Rule of Two plus filesystem/egress deny rules.
Size or audit safety stock with assumption checks - the z*sigma*sqrt(LT) formula plus an empirical stress test of what it actually delivers (cycle service vs fill rate) per…
Audit OSHA training compliance, certification expirations, competency tracking, and LMS integration. Use when you need to evaluate safety training programs, check ANSI Z490.1…
Gate irreversible actions (production deploys, data deletion, moving money, flashing firmware) behind a human-held confirmation token the agent cannot self-issue.
Analyze bash commands for safety risks before execution. Use when user asks about command safety or when reviewing dangerous operations.
Automate Safetyculture tasks via Rube MCP (Composio). Always search tools first for current schemas. — from general/automation