关于 fullstack-debugger
fullstack-debugger 是面向使用 Next.js 15、Cloudflare Workers、Supabase 和边缘部署的开发者的调试技能。它按证据逐层定位故障,并指导验证修复结果。
主要功能
- 01按复现到验证的流程排查
- 02诊断浏览器、服务端和数据库
- 03提供构建与类型检查命令
- 04排查 RLS、CORS、认证和缓存
- 05报告根因并验证修复
适用场景
- 01调查 Next.js 构建失败
- 02诊断 Supabase RLS 或认证问题
- 03检查 Cloudflare Worker 健康状态与 CORS
- 04追踪水合、缓存或部署故障
安装
复制下面的命令,在终端中运行。
$
npx skills add https:/ / github.com/ curiositech/ some_claude_skills --skill fullstack-debugger安装后在 Claude Code 中输入/fullstack-debugger确认已加载。
包含的文件
除 SKILL.md 外,安装时一起下载的参考资料和脚本。
.claude-plugin/plugin.jsonreferences/debug-decision-tree.mderror-patterns.mdscripts/check-cors.shcheck-rls.jsdiagnose.shinspect-cache.sh详细介绍
全栈调试器
适用于现代 Web 技术栈的专家级调试器:Next.js 15、Cloudflare Workers、Supabase 和边缘部署。采用系统化、以证据为依据的故障排查方式。
激活触发条件
在以下情况下激活:“debug”、“not working”、“broken”、“error”、“500 error”、“401”、“403”、“cache issue”、“CORS error”、“RLS policy”、“auth not working”、“blank page”、“hydration error”、“build failed”、“worker not responding”
**不适用于:**功能开发 → 语言技能 | 架构设计 → system-architect | 性能优化 → performance-engineer
调试理念
1. REPRODUCE → Can you make it fail consistently?
2. ISOLATE → Which layer is broken?
3. EVIDENCE → What do logs/network/state show?
4. HYPOTHESIZE → What could cause this?
5. TEST → Validate one hypothesis at a time
6. FIX → Minimal change that resolves issue
7. VERIFY → Confirm fix doesn't break other things
架构层
┌─────────────────────────────────────────────────────────────┐
│ DEBUGGING LAYERS │
├─────────────────────────────────────────────────────────────┤
│ │
│ Layer 1: Browser/Client │
│ ├── Console errors, network tab, React DevTools │
│ ├── localStorage/sessionStorage state │
│ └── React Query cache state │
│ │
│ Layer 2: Next.js Application │
│ ├── Server components vs client components │
│ ├── Build output and static generation │
│ ├── API routes (if any) │
│ └── Hydration mismatches │
│ │
│ Layer 3: Cloudflare Workers │
│ ├── Worker logs (wrangler tail) │
│ ├── KV cache state │
│ ├── CORS headers │
│ └── Rate limiting │
│ │
│ Layer 4: Supabase │
│ ├── Auth state and JWT tokens │
│ ├── RLS policies (most common issue!) │
│ ├── Database queries and indexes │
│ └── Realtime subscriptions │
│ │
│ Layer 5: External APIs │
│ ├── Third-party service availability │
│ ├── API rate limits │
│ └── Response format changes │
│ │
└─────────────────────────────────────────────────────────────┘
快速诊断命令
一次检查所有内容
# Run from next-app/ directory
echo "=== Build Check ===" && npm run build 2>&1 | tail -20
echo "=== TypeScript ===" && npx tsc --noEmit 2>&1 | head -20
echo "=== Lint ===" && npm run lint 2>&1 | head -20
echo "=== Git Status ===" && git status --short
Supabase RLS 诊断
# Check if RLS is blocking queries (most common issue!)
node -e "
const { createClient } = require('@supabase/supabase-js');
const supabase = createClient(
'YOUR_SUPABASE_URL',
'YOUR_ANON_KEY'
);
async function diagnose() {
// Test as anonymous user
const { data, error, count } = await supabase
.from('YOUR_TABLE')
.select('*', { count: 'exact' })
.limit(5);
console.log('Error:', error);
console.log('Count:', count);
console.log('Sample:', data);
}
diagnose();
"
Worker 健康检查
# Check if workers are responding
curl -s -o /dev/null -w "%{http_code}" https://YOUR-WORKER.workers.dev/health
# Check CORS headers
curl -s -D - -o /dev/null -H "Origin: https://yoursite.com" \
https://YOUR-WORKER.workers.dev/api/endpoint | grep -iE "(access-control|x-)"
# Stream worker logs
cd workers/your-worker && npx wrangler tail
缓存检查
# Check Cloudflare KV cache
npx wrangler kv:key list --namespace-id=YOUR_NAMESPACE_ID | head -20
# Get specific cached value
npx wrangler kv:key get --namespace-id=YOUR_NAMESPACE_ID "cache:key"
# Clear a cached item
npx wrangler kv:key delete --namespace-id=YOUR_NAMESPACE_ID "cache:key"
常见问题及解决方案
1. RLS 策略阻止数据访问(最常见!)
症状:
- 查询返回空数组,但没有错误
- 在 Supabase 控制台中正常,但在应用中不行
- 某些用户可以使用,其他用户不行
诊断:
-- In Supabase SQL Editor
-- Check what policies exist
SELECT schemaname, tablename, policyname, permissive, roles, cmd, qual
FROM pg_policies
WHERE tablename = 'your_table';
-- Test as anonymous user
SET ROLE anon;
SELECT * FROM your_table LIMIT 5;
RESET ROLE;
-- Test as authenticated user
SET ROLE authenticated;
SET request.jwt.claims = '{"sub": "user-uuid-here"}';
SELECT * FROM your_table LIMIT 5;
RESET ROLE;
常见修复方法:
-- Allow public read access
CREATE POLICY "Allow public read" ON your_table
FOR SELECT USING (true);
-- Allow authenticated users to read
CREATE POLICY "Allow authenticated read" ON your_table
FOR SELECT TO authenticated USING (true);
-- Allow users to read their own data
CREATE POLICY "Users read own data" ON your_table
FOR SELECT USING (auth.uid() = user_id);
2. CORS 错误
症状:
- “Access to fetch blocked by CORS policy”
- 在 Postman 中正常,但在浏览器中不行
- 预检请求失败
诊断:
# Check what CORS headers are returned
curl -s -D - -o /dev/null \
-H "Origin: https://yoursite.com" \
-H "Access-Control-Request-Method: POST" \
-X OPTIONS \
https://your-worker.workers.dev/api/endpoint
在 Cloudflare Worker 中修复:
// In your worker
const corsHeaders = {
'Access-Control-Allow-Origin': '*', // Or specific domain
'Access-Control-Allow-Methods': 'GET, POST, OPTIONS',
'Access-Control-Allow-Headers': 'Content-Type, Authorization',
};
// Handle preflight
if (request.method === 'OPTIONS') {
return new Response(null, { headers: corsHeaders });
}
// Add to all responses
return new Response(data, {
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
});
3. Auth 状态未持久化
症状:
- 用户已登录,但刷新后显示为已登出
- 本地 Auth 正常,但在生产环境中不行
- 会话随机消失
诊断:
// In browser console
console.log('Session:', await supabase.auth.getSession());
console.log('User:', await supabase.auth.getUser());
console.log('LocalStorage:', Object.keys(localStorage).filter(k => k.includes('supabase')));
常见修复方法:
- 检查 Supabase URL 是否匹配(http 与 https、末尾斜杠)
- 确认 Supabase Auth 设置中的站点 URL
- 检查 Cookie 是否被阻止(Safari、无痕模式)
- 确保 AuthContext 包裹了所有需要 Auth 的组件
4. 水合不匹配
症状:
- “Hydration failed because the initial UI does not match”
- 页面加载时内容闪烁
- 服务端与客户端的内容不同
诊断:
// Temporarily add to suspect component
useEffect(() => {
console.log('Client render:', document.body.innerHTML.slice(0, 500));
}, []);
常见修复方法:
// Use client-only rendering for dynamic content
'use client';
import { useState, useEffect } from 'react';
function DynamicContent() {
const [mounted, setMounted] = useState(false);
useEffect(() => setMounted(true), []);
if (!mounted) return null; // or skeleton
return <div>{/* dynamic content */}</div>;
}
5. Worker 未部署
症状:
- 部署命令成功,但变更未生效
- 仍在运行旧代码
- 新旧行为间歇性交替出现
诊断:
# Check deployment status
npx wrangler deployments list
# View current worker code
npx wrangler deployments view
# Check for multiple environments
npx wrangler whoami
修复方法:
# Force redeploy
npx wrangler deploy --force
# Clear Cloudflare cache
curl -X POST "https://api.cloudflare.com/client/v4/zones/ZONE_ID/purge_cache" \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
--data '{"purge_everything":true}'
6. TypeScript 缓存幽灵问题
症状:
- 错误引用已删除或已更改的代码
- 类型与当前代码不匹配
- 现有文件却出现“Cannot find module”错误
修复方法:
# Nuclear option - clear all caches
rm -rf .next node_modules/.cache tsconfig.tsbuildinfo
npm run build
# Or just TypeScript cache
rm -rf node_modules/.cache/typescript
npx tsc --build --clean
7. 静态导出问题
症状:
- “Error: Page X couldn't be rendered statically”
- 动态路由在静态导出时失败
- 部署后 API 路由无法使用
诊断:
# Check next.config for output mode
grep -A5 "output:" next.config.ts
# Find dynamic components
grep -r "useSearchParams\|usePathname\|cookies()\|headers()" src/
修复方法:
// For components using dynamic APIs
export const dynamic = 'force-dynamic';
// or wrap in Suspense with fallback
// For generateStaticParams
export async function generateStaticParams() {
return [{ slug: 'page1' }, { slug: 'page2' }];
}
8. 速率限制问题
症状:
- 多次请求后出现 429 错误
- 一开始正常,之后停止工作
- 不同 IP 的行为不同
诊断:
# Check rate limit headers
curl -i https://your-worker.workers.dev/api/endpoint 2>&1 | grep -i ratelimit
# Check KV for rate limit keys
npx wrangler kv:key list --namespace-id=RATE_LIMIT_KV_ID | grep rate
修复方法:
# Clear rate limit for an IP
npx wrangler kv:key delete --namespace-id=RATE_LIMIT_KV_ID "rate:192.168.1.1"
# Adjust limits in wrangler.toml
RATE_LIMIT_REQUESTS = "100"
RATE_LIMIT_WINDOW = "3600"
9. 会议/地点数据问题
症状:
- 某些区域找不到会议
- 会议数据过时
- 缓存显示错误数据
诊断:
# Check cache status for a location
curl -s -D - -o /dev/null \
-H "Origin: https://yoursite.com" \
"https://your-proxy.workers.dev/api/all?lat=45.52&lng=-122.68&radius=25" \
| grep -iE "(x-cache|x-geohash|x-source)"
# Force cache refresh
curl -H "Origin: https://yoursite.com" \
"https://your-proxy.workers.dev/warm"
# Check Supabase for meeting count
node -e "
const { createClient } = require('@supabase/supabase-js');
const supabase = createClient('URL', 'KEY');
supabase.from('meetings').select('*', { count: 'exact', head: true })
.then(({count}) => console.log('Total meetings:', count));
"
10. Cloudflare Pages 构建失败
症状:
- 本地正常,但部署失败
- 出现“Module not found”错误
- 内存超限
诊断:
# Check build output locally
NODE_ENV=production npm run build 2>&1 | tee build.log
# Check for conditional imports
grep -r "require(" src/ --include="*.ts" --include="*.tsx"
# Check bundle size
npx next-bundle-analyzer
修复方法:
// next.config.ts - increase memory
module.exports = {
experimental: {
memoryBasedWorkersCount: true,
},
// Reduce bundle size
webpack: (config) => {
config.externals = [...(config.externals || []), 'sharp'];
return config;
}
};
调试脚本
scripts/diagnose.sh
#!/bin/bash
# Run all diagnostics
echo "=== Environment ==="
node -v && npm -v
echo "=== Dependencies ==="
npm ls --depth=0 2>&1 | grep -E "(UNMET|missing)"
echo "=== TypeScript ==="
npx tsc --noEmit 2>&1 | head -30
echo "=== Build ==="
npm run build 2>&1 | tail -30
echo "=== Workers ==="
for worker in workers/*/; do
echo "Worker: $worker"
(cd "$worker" && npx wrangler whoami 2>/dev/null)
done
echo "=== Supabase ==="
npx supabase status 2>/dev/null || echo "Supabase CLI not configured"
scripts/check-rls.js
// Check RLS policies are working correctly
const { createClient } = require('@supabase/supabase-js');
const supabase = createClient(
process.env.NEXT_PUBLIC_SUPABASE_URL,
process.env.NEXT_PUBLIC_SUPABASE_ANON_KEY
);
async function checkTable(table) {
console.log(`\n=== Checking ${table} ===`);
const { data, error, count } = await supabase
.from(table)
.select('*', { count: 'exact' })
.limit(1);
if (error) {
console.log(`ERROR: ${error.message}`);
} else {
console.log(`OK: ${count} rows accessible`);
}
}
// Check critical tables
['profiles', 'meetings', 'forum_posts', 'journal_entries'].forEach(checkTable);
验证清单
[ ] Can reproduce the issue consistently
[ ] Identified which layer is failing (client/Next/Worker/Supabase/API)
[ ] Checked browser console for errors
[ ] Checked network tab for failed requests
[ ] Checked worker logs (wrangler tail)
[ ] Verified RLS policies allow access
[ ] Tested with fresh browser/incognito
[ ] Cleared all caches (browser, React Query, KV, TS)
[ ] Checked environment variables match production
[ ] Verified CORS headers are correct
[ ] Tested on production URL (not just localhost)
[ ] Created minimal reproduction case
输出
调试时,请始终提供:
- 根本原因——具体出了什么问题
- 证据——能够证明问题的日志、错误或查询
- 修复方法——解决问题所需的最小代码变更
- 验证方式——如何确认问题已修复
- 预防措施——如何避免今后再次发生
可用工具
Read、Write、Edit——文件操作Bash——运行命令、curl、wranglerGrep、Glob——搜索代码库WebFetch——测试端点mcp__supabase__*——直接执行 Supabase 操作mcp__playwright__*——用于 UI 测试的浏览器自动化
评论
支持 Markdown