一键在 Manus 中运行任何 Skill

revealui-safety

星标6

分支0

更新时间2026年7月16日 20:27

RevealUI safety guardrails for any code task — editing, writing, creating, fixing, refactoring, changing, adding, updating, or removing files. Protects credentials, enforces import boundaries, ensures code quality, and verifies work before completion.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

RevealUIStudio

RevealUIStudio/revealui

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

RevealUI Safety

Follow these rules for ALL code changes in the RevealUI monorepo.

Protected Files — Ask Before Editing

.env* files (.env, .env.local, .env.production, etc.)
Lock files: pnpm-lock.yaml, package-lock.json, yarn.lock
Database schema files in packages/db/src/schema/ — changes require migration planning

Protected Paths — Never Edit

/mnt/c/, /mnt/e/ — Windows mounts (read-only)
System/credential directories: /etc/, ~/.ssh/, ~/.gnupg/, ~/.aws/

Import Boundaries

All persistence goes through the single Drizzle/Neon client (@revealui/db). Do not introduce a second database client or a separate vector/auth SDK — vector data lives in pgvector on the same Neon database. (A customer-facing Supabase MCP adapter exists for connecting a customer's own Supabase project; that is not an internal store.)

Code Quality

Never use any — use unknown + type guards
Never add console.* in production code — use @revealui/utils logger
Never hardcode API keys, tokens, passwords, or secrets
Use crypto.randomInt() for security-sensitive values, not Math.random()

Static Analysis

For security and architecture validation scripts, prefer AST-based analysis over regex when the rule depends on syntax or code shape
Use regex only for heuristic inventory scans (for example obvious secret patterns), not as the source of truth for code-security conclusions

After Every Edit

Run npx biome check --write <file> on each file you edit before moving on.

Before Claiming Done

Run pnpm gate:quick and confirm no new errors
Review git diff for unintended changes
Ensure conventional commit format: type(scope): description
Git identity: commit as your own GitHub noreply address (<ID>+<username>@users.noreply.github.com)

Known Limitation

These rules are advisory. Unlike Claude Code (which enforces via lifecycle hooks), Codex has no hook system. If working on sensitive files, explicitly invoke $revealui-safety to load these rules.

同仓库更多 Skills

同仓库

revealui-conventions

RevealUIStudio/revealui

RevealUI coding conventions for any code task — writing, editing, reviewing, creating, fixing, refactoring, changing, adding, or updating TypeScript, React, CSS, or config files. Covers TypeScript strict mode, ES Modules, Biome formatting, Tailwind v4 syntax, conventional commits, monorepo workspace protocol, feature gating, parameterization, and unused declaration policy.

2026-07-166

revealui-db

RevealUIStudio/revealui

RevealUI database conventions for any task involving database, schema, query, migration, Drizzle ORM, Neon, PostgreSQL, pgvector, vectors, embeddings, or data modeling. Covers the single Neon-primary database and its pgvector tables.

2026-07-166

revealui-debugging

RevealUIStudio/revealui

Systematic debugging workflow for RevealUI. Use when encountering any bug, test failure, unexpected behavior, error, or broken functionality. Prevents shotgun debugging.

2026-07-166

revealui-deploy

RevealUIStudio/revealui

RevealUI deployment guide — Vercel configuration, GitHub Actions deploy workflow, secret management, domain aliasing, and troubleshooting. Invoke when working on deploy.yml, vercel.json, deployment secrets, domain configuration, or debugging deploy failures.

2026-07-166

revealui-review

RevealUIStudio/revealui

Code review checklist for RevealUI. Use when reviewing code, completing a feature, checking quality, or before committing. Invoke explicitly with $revealui-review.

2026-07-166

revealui-tdd

RevealUIStudio/revealui

Test-driven development workflow for RevealUI. Use when implementing any feature, fixing bugs, adding functionality, or writing new code. Enforces write-test-first, red-green-refactor cycle. Works with Vitest and React Testing Library.

2026-07-166

name	revealui-safety
description	RevealUI safety guardrails for any code task — editing, writing, creating, fixing, refactoring, changing, adding, updating, or removing files. Protects credentials, enforces import boundaries, ensures code quality, and verifies work before completion.

RevealUI Safety

Follow these rules for ALL code changes in the RevealUI monorepo.

Protected Files — Ask Before Editing

.env* files (.env, .env.local, .env.production, etc.)
Lock files: pnpm-lock.yaml, package-lock.json, yarn.lock
Database schema files in packages/db/src/schema/ — changes require migration planning

Protected Paths — Never Edit

/mnt/c/, /mnt/e/ — Windows mounts (read-only)
System/credential directories: /etc/, ~/.ssh/, ~/.gnupg/, ~/.aws/

Import Boundaries

Code Quality

Never use any — use unknown + type guards
Never add console.* in production code — use @revealui/utils logger
Never hardcode API keys, tokens, passwords, or secrets
Use crypto.randomInt() for security-sensitive values, not Math.random()

Static Analysis

For security and architecture validation scripts, prefer AST-based analysis over regex when the rule depends on syntax or code shape
Use regex only for heuristic inventory scans (for example obvious secret patterns), not as the source of truth for code-security conclusions

After Every Edit

Run npx biome check --write <file> on each file you edit before moving on.

Before Claiming Done

Run pnpm gate:quick and confirm no new errors
Review git diff for unintended changes
Ensure conventional commit format: type(scope): description
Git identity: commit as your own GitHub noreply address (<ID>+<username>@users.noreply.github.com)

Known Limitation

These rules are advisory. Unlike Claude Code (which enforces via lifecycle hooks), Codex has no hook system. If working on sensitive files, explicitly invoke $revealui-safety to load these rules.