Initial commit

This commit is contained in:
Zhongwei Li
2025-11-30 08:24:03 +08:00
commit d3ec204941
27 changed files with 4067 additions and 0 deletions

642
references/common-errors.md Normal file
View File

@@ -0,0 +1,642 @@
# Clerk Authentication - Common Errors & Solutions
**Last Updated**: 2025-10-22
This document provides detailed troubleshooting for all 10 documented Clerk authentication issues.
---
## Error #1: Missing Clerk Secret Key
### Symptoms
\`\`\`
Error: Missing Clerk Secret Key or API Key
\`\`\`
### Why It Happens
- \`CLERK_SECRET_KEY\` environment variable not set
- Environment file not loaded
- Wrong environment file name
### Solutions
**Next.js**:
1. Create \`.env.local\` in project root
2. Add: \`CLERK_SECRET_KEY=sk_test_...\`
3. Restart dev server
**Cloudflare Workers**:
1. Local: Create \`.dev.vars\` with \`CLERK_SECRET_KEY=sk_test_...\`
2. Production: Run \`wrangler secret put CLERK_SECRET_KEY\`
### Prevention
- Always set secret key before running app
- Verify with: \`echo $CLERK_SECRET_KEY\` (should be empty - it's private!)
- Check environment is loading: add \`console.log(!!process.env.CLERK_SECRET_KEY)\`
**Source**: https://stackoverflow.com/questions/77620604
---
## Error #2: API Key → Secret Key Migration (Core 2)
### Symptoms
\`\`\`
Warning: apiKey is deprecated, use secretKey instead
TypeError: Cannot read properties of undefined
\`\`\`
### Why It Happens
- Upgrading from @clerk/backend v1 to v2 (Core 2)
- Old code uses \`apiKey\` parameter
- Breaking change in Core 2
### Solutions
**Before (v1)**:
\`\`\`typescript
const clerk = createClerkClient({
apiKey: process.env.CLERK_API_KEY // ❌ Deprecated
})
\`\`\`
**After (v2)**:
\`\`\`typescript
const clerk = createClerkClient({
secretKey: process.env.CLERK_SECRET_KEY // ✅ Correct
})
\`\`\`
### Prevention
- Use \`secretKey\` in all new code
- Search codebase for \`apiKey:\` and replace with \`secretKey:\`
- Update environment variable name from \`CLERK_API_KEY\` to \`CLERK_SECRET_KEY\`
**Source**: https://clerk.com/docs/upgrade-guides/core-2/backend
---
## Error #3: JWKS Cache Race Condition
### Symptoms
\`\`\`
Error: No JWK available
Token verification fails intermittently
\`\`\`
### Why It Happens
- Race condition in older @clerk/backend versions
- JWKS cache not populated before verification
- Fixed in recent versions
### Solutions
**Update Package**:
\`\`\`bash
npm install @clerk/backend@latest
# Ensure version is 2.17.2 or later
\`\`\`
**Verify Version**:
\`\`\`bash
npm list @clerk/backend
\`\`\`
### Prevention
- Always use latest stable @clerk/backend version
- This issue is fixed in modern versions
**Source**: https://github.com/clerk/javascript/blob/main/packages/backend/CHANGELOG.md
---
## Error #4: Missing authorizedParties (CSRF Vulnerability)
### Symptoms
- No error, but security vulnerability
- Tokens from other domains accepted
- CSRF attacks possible
### Why It Happens
- \`authorizedParties\` not set in \`verifyToken()\`
- Clerk accepts tokens from any domain by default
### Solutions
**Always Set authorizedParties**:
\`\`\`typescript
const { data, error } = await verifyToken(token, {
secretKey: env.CLERK_SECRET_KEY,
authorizedParties: [
'http://localhost:5173', // Development
'https://yourdomain.com', // Production
], // ✅ Required for security
})
\`\`\`
### Prevention
- Always set \`authorizedParties\` in production
- List all domains that should be able to make requests
- Never use wildcard or leave empty
**Source**: https://clerk.com/docs/reference/backend/verify-token
---
## Error #5: Import Path Changes (Core 2 Upgrade)
### Symptoms
\`\`\`
Error: Cannot find module '@clerk/backend'
Module not found: Can't resolve '@clerk/backend/errors'
\`\`\`
### Why It Happens
- Import paths changed in Core 2
- Old imports don't work in new version
### Solutions
**Update Imports**:
\`\`\`typescript
// Before (v1)
import { TokenVerificationError } from '@clerk/backend'
// After (v2)
import { TokenVerificationError } from '@clerk/backend/errors'
\`\`\`
### Common Path Changes
| Old Path | New Path |
|----------|----------|
| \`@clerk/backend\` (errors) | \`@clerk/backend/errors\` |
| \`@clerk/backend\` (types) | \`@clerk/backend/types\` |
### Prevention
- Follow official migration guide
- Update all imports when upgrading
**Source**: https://clerk.com/docs/upgrade-guides/core-2/backend
---
## Error #6: JWT Size Limit Exceeded
### Symptoms
\`\`\`
Error: Token exceeds maximum size
Session token too large
\`\`\`
### Why It Happens
- Custom JWT claims exceed 1.2KB limit
- Too much data in \`publicMetadata\`
- Clerk's default claims + your claims > 1.2KB
### Solutions
**Reduce Custom Claims**:
\`\`\`json
{
"email": "{{user.primary_email_address}}",
"role": "{{user.public_metadata.role}}"
}
\`\`\`
**Store Large Data in Database**:
- Use JWT for minimal claims (user ID, role)
- Store profile data, preferences in your database
- Fetch after authentication
### Prevention
- Keep custom claims minimal
- Use database for large/complex data
- Monitor token size during development
**Source**: https://clerk.com/docs/backend-requests/making/custom-session-token
---
## Error #7: Deprecated API Version v1
### Symptoms
\`\`\`
Warning: API version v1 is deprecated
Please upgrade to API version 2025-04-10
\`\`\`
### Why It Happens
- Using old SDK version
- API version 1 deprecated April 2025
### Solutions
**Update SDKs**:
\`\`\`bash
npm install @clerk/nextjs@latest
npm install @clerk/backend@latest
npm install @clerk/clerk-react@latest
\`\`\`
**Verify Versions**:
- @clerk/nextjs: 6.0.0+
- @clerk/backend: 2.0.0+
- @clerk/clerk-react: 5.0.0+
### Prevention
- Keep SDKs updated
- Use latest stable versions
- Follow deprecation warnings
**Source**: https://clerk.com/docs/upgrade-guides/core-2/backend
---
## Error #8: ClerkProvider JSX Component Error
### Symptoms
\`\`\`
Error: 'ClerkProvider' cannot be used as a JSX component
Its element type 'ReactElement<any, any> | Component<...>' is not a valid JSX element
\`\`\`
### Why It Happens
- React version mismatch
- @clerk/clerk-react not compatible with React version
- TypeScript type conflicts
### Solutions
**Update Packages**:
\`\`\`bash
npm install @clerk/clerk-react@latest react@latest react-dom@latest
\`\`\`
**Ensure Compatibility**:
- React 19: Use @clerk/clerk-react@5.51.0+
- React 18: Use @clerk/clerk-react@5.0.0+
### Prevention
- Keep React and Clerk packages in sync
- Use compatible versions
**Source**: https://stackoverflow.com/questions/79265537
---
## Error #9: Async auth() Helper Confusion
### Symptoms
\`\`\`
Error: auth() is not a function
TypeError: Cannot read properties of undefined
\`\`\`
### Why It Happens
- @clerk/nextjs v6 made \`auth()\` async (breaking change)
- Old code doesn't await \`auth()\`
### Solutions
**Before (v5)**:
\`\`\`typescript
const { userId } = auth() // ❌ Sync in v5
\`\`\`
**After (v6)**:
\`\`\`typescript
const { userId } = await auth() // ✅ Async in v6
\`\`\`
### All Affected Code
- \`const { userId } = await auth()\`
- \`const user = await currentUser()\`
- \`await auth.protect()\` (in middleware)
### Prevention
- Always await auth-related functions in v6+
- Update all instances when upgrading
**Source**: https://clerk.com/changelog/2024-10-22-clerk-nextjs-v6
---
## Error #10: Environment Variable Misconfiguration
### Symptoms
\`\`\`
Error: Missing Publishable Key
Warning: Secret key exposed to client
\`\`\`
### Why It Happens
- Wrong environment variable prefix
- Secrets committed to version control
- Environment file not loaded
### Solutions
**Next.js Prefix Rules**:
- Client variables: \`NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY\`
- Server variables: \`CLERK_SECRET_KEY\` (no prefix!)
**Vite Prefix Rules**:
- Client variables: \`VITE_CLERK_PUBLISHABLE_KEY\`
**Cloudflare Workers**:
- Local: \`.dev.vars\` file
- Production: \`wrangler secret put CLERK_SECRET_KEY\`
### Prevention Checklist
- [ ] Use correct prefix for framework
- [ ] Never use \`NEXT_PUBLIC_\` or \`VITE_\` for secrets
- [ ] Add \`.env.local\` and \`.dev.vars\` to \`.gitignore\`
- [ ] Use different keys for dev and production
- [ ] Restart dev server after changing env vars
**Source**: General best practices
---
## Error #11: 431 Request Header Fields Too Large (Vite Dev Mode)
### Symptoms
```
431 Request Header Fields Too Large
Failed to load resource: the server responded with a status of 431 ()
```
Happens when:
- Signing in to your Vite app in development mode
- After adding custom JWT claims
- URL contains very long `__clerk_handshake=...` parameter
### Why It Happens
- Clerk's authentication handshake passes a JWT token in the URL as `__clerk_handshake=...`
- Vite dev server (Node.js http) has default 8KB request header limit
- Clerk development tokens are larger than production tokens
- Custom JWT claims increase handshake token size
- Long URLs with JWT tokens exceed Vite's header limit
### Solutions
**Solution 1: Increase Node.js Header Limit (Recommended)**
Update `package.json`:
```json
{
"scripts": {
"dev": "NODE_OPTIONS='--max-http-header-size=32768' vite",
"build": "vite build"
}
}
```
This increases the limit from 8KB to 32KB.
**For Windows PowerShell**:
```json
{
"scripts": {
"dev": "cross-env NODE_OPTIONS=--max-http-header-size=32768 vite"
}
}
```
Install `cross-env`: `npm install -D cross-env`
**Solution 2: Temporary Workaround**
Clear browser state:
1. Open DevTools (F12)
2. Right-click refresh button → "Empty Cache and Hard Reload"
3. Or: Application tab → Clear all storage
4. Sign out and sign back in
This removes the problematic handshake token and forces Clerk to create a fresh one.
### What DOESN'T Work
**This won't fix it**:
```typescript
// vite.config.ts
export default defineConfig({
server: {
headers: {
'Cache-Control': 'no-cache', // Wrong - this sets RESPONSE headers
},
},
})
```
The error is about REQUEST headers (incoming), not RESPONSE headers (outgoing).
### Prevention
- Expect this error when testing custom JWT claims in dev mode
- Set `NODE_OPTIONS` in `package.json` from the start
- Keep custom claims minimal during development
- Production deployments (Cloudflare Workers, Vercel) don't have this issue
### Difference from Error #6
**Error #6** (JWT Size Limit Exceeded):
- About session token size in cookies
- 1.2KB limit for custom claims
- Affects production
- Browser cookie size limit
**Error #11** (431 Header Error):
- About handshake token size in URL parameters
- 8KB default limit (increase to 32KB)
- Only affects Vite dev mode
- Node.js HTTP server limit
### Why Production Isn't Affected
- Production builds don't use Vite dev server
- Cloudflare Workers, Vercel, Netlify have higher header limits
- Production tokens are smaller (no dev overhead)
- Handshake flow is optimized in production
**Source**: Real-world developer experience with Vite + Clerk + Custom JWT claims
---
## Quick Debugging Checklist
When auth isn't working:
1. **Check Environment Variables**
- [ ] \`CLERK_SECRET_KEY\` set?
- [ ] \`CLERK_PUBLISHABLE_KEY\` or \`NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY\` set?
- [ ] Correct prefix for framework?
- [ ] Dev server restarted after changes?
2. **Check Package Versions**
- [ ] Using latest stable versions?
- [ ] React version compatible with Clerk?
- [ ] All Clerk packages same major version?
3. **Check Code Patterns**
- [ ] \`auth()\` being awaited? (v6+)
- [ ] \`authorizedParties\` set in \`verifyToken()\`?
- [ ] \`isLoaded\` checked before rendering?
- [ ] Using \`secretKey\` not \`apiKey\`?
4. **Check Network**
- [ ] Token in \`Authorization: Bearer <token>\` header?
- [ ] CORS configured if API is different domain?
- [ ] Clerk Dashboard shows API requests?
5. **Check Clerk Dashboard**
- [ ] Application configured correctly?
- [ ] Development/production keys match environment?
- [ ] Custom JWT template valid?
- [ ] Webhooks configured if using?
---
**Still Having Issues?**
1. Check official Clerk docs: https://clerk.com/docs
2. Search Clerk Discord: https://clerk.com/discord
3. File GitHub issue: https://github.com/clerk/javascript/issues
4. Check Clerk status: https://status.clerk.com
---
## Error #12: Deprecated Redirect URL Props
### Symptoms
```
Clerk: The prop "afterSignInUrl" is deprecated and should be replaced with the new "fallbackRedirectUrl" or "forceRedirectUrl" props instead.
Clerk: The prop "afterSignUpUrl" is deprecated and should be replaced with the new "fallbackRedirectUrl" or "forceRedirectUrl" props instead.
```
Appears in:
- Browser console
- Development mode
- When using `<SignIn>` or `<SignUp>` components from `@clerk/clerk-react`
### Why It Happens
- Clerk updated redirect prop naming in v5.x for clarity
- Old props: `afterSignInUrl`, `afterSignUpUrl`
- New props: `fallbackRedirectUrl`, `forceRedirectUrl`
- Old props still work but trigger deprecation warnings
### Solution
**Replace deprecated props:**
**Old (Deprecated)**:
```tsx
<SignIn
afterSignInUrl="/"
signUpUrl="/signup"
/>
<SignUp
afterSignUpUrl="/"
signInUrl="/login"
/>
```
**New (Recommended)**:
```tsx
<SignIn
fallbackRedirectUrl="/"
signUpUrl="/signup"
/>
<SignUp
fallbackRedirectUrl="/"
signInUrl="/login"
/>
```
### Choosing Between fallbackRedirectUrl and forceRedirectUrl
**Use `fallbackRedirectUrl`** (Most Common):
- Used as a fallback if no redirect URL is in query params
- Allows Clerk to use `redirect_url` from query string first
- More flexible for return-to-previous-page flows
- **Recommended for most use cases**
**Use `forceRedirectUrl`** (Rare):
- Always redirects here, ignoring query params
- Used when you want strict control
- Overrides any `redirect_url` in query string
- Use only when you need to force a specific destination
### Example: Complete Migration
```tsx
// LoginPage.tsx
import { SignIn } from '@clerk/clerk-react';
export function LoginPage() {
return (
<SignIn
routing="path"
path="/login"
signUpUrl="/signup"
fallbackRedirectUrl="/" // ✅ Use this instead of afterSignInUrl
appearance={{
elements: {
rootBox: 'mx-auto',
card: 'shadow-lg',
},
}}
/>
);
}
```
```tsx
// SignupPage.tsx
import { SignUp } from '@clerk/clerk-react';
export function SignupPage() {
return (
<SignUp
routing="path"
path="/signup"
signInUrl="/login"
fallbackRedirectUrl="/" // ✅ Use this instead of afterSignUpUrl
appearance={{
elements: {
rootBox: 'mx-auto',
card: 'shadow-lg',
},
}}
/>
);
}
```
### Migration Checklist
- [ ] Search codebase for `afterSignInUrl`
- [ ] Replace with `fallbackRedirectUrl`
- [ ] Search codebase for `afterSignUpUrl`
- [ ] Replace with `fallbackRedirectUrl`
- [ ] Test redirect flow after sign in
- [ ] Test redirect flow after sign up
- [ ] Verify no console warnings
### Why This Matters
**Clarity**: The new naming makes it clear these are fallback/default redirects
**Future-Proofing**: Old props may be removed in future major versions
**Best Practices**: Using current APIs ensures compatibility with new features
**Source**: Clerk v5.x Changelog & Official Migration Guide
**Reference**: https://clerk.com/docs/guides/custom-redirects#redirect-url-props
---
**Error #12** (Deprecated Redirect Props):
- About redirect URL prop naming
- Simple find-replace fix
- Use `fallbackRedirectUrl` for most cases
- Only affects React-based projects

View File

@@ -0,0 +1,26 @@
# [TODO: Reference Document Name]
[TODO: This file contains reference documentation that Claude can load when needed.]
[TODO: Delete this file if you don't have reference documentation to provide.]
## Purpose
[TODO: Explain what information this document contains]
## When Claude Should Use This
[TODO: Describe specific scenarios where Claude should load this reference]
## Content
[TODO: Add your reference content here - schemas, guides, specifications, etc.]
---
**Note**: This file is NOT loaded into context by default. Claude will only load it when:
- It determines the information is needed
- You explicitly ask Claude to reference it
- The SKILL.md instructions direct Claude to read it
Keep this file under 10k words for best performance.

View File

@@ -0,0 +1,660 @@
# Clerk JWT Claims: Complete Reference
**Last Updated**: 2025-10-22
**Clerk API Version**: 2025-04-10
**Status**: Production Ready ✅
---
## Overview
This guide provides comprehensive documentation for all JWT claims available in Clerk, including default claims, user property shortcodes, organization claims, metadata access patterns, and advanced template features.
**Use this guide when**:
- Creating custom JWT templates in Clerk Dashboard
- Integrating with third-party services requiring specific token formats
- Implementing role-based access control (RBAC)
- Building multi-tenant applications
- Accessing user data in backend services
---
## Table of Contents
1. [Default Claims (Auto-Included)](#default-claims-auto-included)
2. [User Property Shortcodes](#user-property-shortcodes)
3. [Organization Claims](#organization-claims)
4. [Metadata Access](#metadata-access)
5. [Advanced Template Features](#advanced-template-features)
6. [Creating JWT Templates](#creating-jwt-templates)
7. [TypeScript Type Safety](#typescript-type-safety)
8. [Common Use Cases](#common-use-cases)
9. [Limitations & Gotchas](#limitations--gotchas)
10. [Official Documentation](#official-documentation)
---
## Default Claims (Auto-Included)
Every JWT generated by Clerk automatically includes these claims. **These cannot be overridden in custom templates**.
```json
{
"azp": "http://localhost:3000", // Authorized party (your app URL)
"exp": 1639398300, // Expiration time (Unix timestamp)
"iat": 1639398272, // Issued at (Unix timestamp)
"iss": "https://your-app.clerk.accounts.dev", // Issuer (Clerk instance URL)
"jti": "10db7f531a90cb2faea4", // JWT ID (unique identifier)
"nbf": 1639398220, // Not before (Unix timestamp)
"sub": "user_1deJLArSTiWiF1YdsEWysnhJLLY" // Subject (user ID)
}
```
### Claim Descriptions
| Claim | Name | Description | Can Override? |
|-------|------|-------------|---------------|
| `azp` | Authorized Party | The URL of your application | ❌ No |
| `exp` | Expiration Time | When the token expires (Unix timestamp) | ❌ No |
| `iat` | Issued At | When the token was created (Unix timestamp) | ❌ No |
| `iss` | Issuer | Your Clerk instance URL | ❌ No |
| `jti` | JWT ID | Unique identifier for this token | ❌ No |
| `nbf` | Not Before | Token not valid before this time (Unix timestamp) | ❌ No |
| `sub` | Subject | User ID (same as `userId` in auth objects) | ❌ No |
**IMPORTANT**: These claims consume approximately **200-300 bytes** of the 4KB cookie limit, leaving ~1.2KB for custom claims.
---
## User Property Shortcodes
Shortcodes are placeholder strings that Clerk replaces with actual user data during token generation. Use double curly braces: `{{shortcode}}`.
### Basic User Properties
```json
{
"user_id": "{{user.id}}", // User's unique ID
"first_name": "{{user.first_name}}", // User's first name (or null)
"last_name": "{{user.last_name}}", // User's last name (or null)
"full_name": "{{user.full_name}}", // User's full name (or null)
"email": "{{user.primary_email_address}}", // Primary email address
"phone": "{{user.primary_phone_address}}", // Primary phone number (or null)
"avatar": "{{user.image_url}}", // User's profile image URL
"created_at": "{{user.created_at}}", // Account creation timestamp
"username": "{{user.username}}", // Username (if enabled)
"email_verified": "{{user.email_verified}}", // Boolean: email verified?
"phone_verified": "{{user.phone_number_verified}}" // Boolean: phone verified?
}
```
### Verification Status
```json
{
"has_verified_email": "{{user.email_verified}}",
"has_verified_phone": "{{user.phone_number_verified}}",
"has_verified_contact": "{{user.email_verified || user.phone_number_verified}}"
}
```
### Complete User Shortcode Reference
| Shortcode | Type | Description | Example Value |
|-----------|------|-------------|---------------|
| `{{user.id}}` | string | User's unique identifier | `"user_2abc..."` |
| `{{user.first_name}}` | string \| null | First name | `"John"` |
| `{{user.last_name}}` | string \| null | Last name | `"Doe"` |
| `{{user.full_name}}` | string \| null | Full name (computed) | `"John Doe"` |
| `{{user.primary_email_address}}` | string | Primary email | `"john@example.com"` |
| `{{user.primary_phone_address}}` | string \| null | Primary phone | `"+1234567890"` |
| `{{user.image_url}}` | string | Profile image URL | `"https://..."` |
| `{{user.created_at}}` | number | Unix timestamp | `1639398272` |
| `{{user.username}}` | string \| null | Username (if enabled) | `"johndoe"` |
| `{{user.email_verified}}` | boolean | Email verified? | `true` |
| `{{user.phone_number_verified}}` | boolean | Phone verified? | `false` |
| `{{user.public_metadata}}` | object | All public metadata | `{...}` |
| `{{user.unsafe_metadata}}` | object | All unsafe metadata | `{...}` |
---
## Organization Claims
Clerk includes claims for the **active organization** (if user is in one and has selected it).
### Active Organization Claims
```json
{
"org_id": "org_2abc...", // Active organization ID
"org_slug": "acme-corp", // Active organization slug
"org_role": "org:admin" // User's role in active organization
}
```
**CRITICAL CHANGE (Core 2)**:
-**New**: `org_id`, `org_slug`, `org_role` (active org only)
-**Removed**: `orgs` claim (previously contained all user organizations)
### Accessing Organization Data in Templates
```json
{
"organization": {
"id": "{{user.public_metadata.org_id}}",
"name": "{{user.public_metadata.org_name}}",
"role": "{{user.public_metadata.org_role}}"
}
}
```
**Note**: For all organizations (not just active), you must:
1. Store org data in `user.public_metadata` via Clerk API
2. Use custom JWT template to include it
---
## Metadata Access
Clerk provides two metadata fields for storing custom user data:
### Public Metadata
- ✅ Accessible on client side
- ✅ Included in user objects
- ✅ Can be included in JWT templates
- ❌ Should NOT contain sensitive data
### Unsafe Metadata
- ⚠️ Name is misleading - it's for "unvalidated" data
- ✅ Accessible on client side
- ✅ Can be included in JWT templates
- ❌ Should NOT contain sensitive data
### Basic Metadata Access
```json
{
"all_public": "{{user.public_metadata}}",
"all_unsafe": "{{user.unsafe_metadata}}"
}
```
**Result**:
```json
{
"all_public": {
"role": "admin",
"department": "engineering"
},
"all_unsafe": {
"onboardingComplete": true
}
}
```
### Nested Metadata with Dot Notation
For nested objects, use dot notation to access specific fields:
**User's public_metadata**:
```json
{
"profile": {
"interests": ["hiking", "knitting"],
"bio": "Software engineer passionate about..."
},
"addresses": {
"Home": "2355 Pointe Lane, 56301 Minnesota",
"Work": "3759 Newton Street, 33487 Florida"
},
"role": "admin",
"department": "engineering"
}
```
**JWT Template**:
```json
{
"role": "{{user.public_metadata.role}}",
"department": "{{user.public_metadata.department}}",
"interests": "{{user.public_metadata.profile.interests}}",
"home_address": "{{user.public_metadata.addresses.Home}}"
}
```
**Generated Token**:
```json
{
"role": "admin",
"department": "engineering",
"interests": ["hiking", "knitting"],
"home_address": "2355 Pointe Lane, 56301 Minnesota"
}
```
**Why This Matters**: Dot notation prevents including the entire metadata object, reducing token size.
---
## Advanced Template Features
### String Interpolation
Combine multiple shortcodes into a single string:
```json
{
"full_name": "{{user.last_name}} {{user.first_name}}",
"greeting": "Hello, {{user.first_name}}!",
"email_with_name": "{{user.full_name}} <{{user.primary_email_address}}>"
}
```
**Result**:
```json
{
"full_name": "Doe John",
"greeting": "Hello, John!",
"email_with_name": "John Doe <john@example.com>"
}
```
**IMPORTANT**: Interpolated values are **always strings**, even if null values are present.
### Conditional Expressions (Fallbacks)
Use `||` operator to provide fallback values:
```json
{
"full_name": "{{user.full_name || 'Guest User'}}",
"age": "{{user.public_metadata.age || 18}}",
"role": "{{user.public_metadata.role || 'user'}}",
"verified": "{{user.email_verified || user.phone_number_verified}}"
}
```
**How It Works**:
- Returns first **non-falsy** operand
- Final operand serves as default
- String literals require **single quotes**: `'default'`
- Can chain multiple fallbacks
**Example with Multiple Fallbacks**:
```json
{
"age": "{{user.public_metadata.age || user.unsafe_metadata.age || 18}}"
}
```
This checks:
1. `user.public_metadata.age` (if null/false, continue)
2. `user.unsafe_metadata.age` (if null/false, continue)
3. `18` (default value)
### Boolean Checks
```json
{
"has_verified_contact": "{{user.email_verified || user.phone_number_verified}}",
"is_complete": "{{user.public_metadata.profileComplete || false}}"
}
```
**Result**:
```json
{
"has_verified_contact": true, // If either email or phone is verified
"is_complete": false // If profileComplete is not set
}
```
---
## Creating JWT Templates
### Step 1: Navigate to Clerk Dashboard
1. Go to **Sessions** page in Clerk Dashboard
2. Click **Customize session token**
3. Choose **Create template** or use pre-built templates
### Step 2: Configure Template Properties
**Template Properties**:
```json
{
"name": "supabase", // Unique identifier (lowercase, no spaces)
"lifetime": 3600, // Token expiration in seconds (default: 60)
"allowed_clock_skew": 5, // Leeway for clock differences (default: 5)
"claims": { // Your custom claims
"email": "{{user.primary_email_address}}",
"role": "{{user.public_metadata.role}}"
}
}
```
### Step 3: Use Template in Code
**React / Next.js**:
```typescript
import { useAuth } from '@clerk/nextjs'
function MyComponent() {
const { getToken } = useAuth()
// Get token with custom template
const token = await getToken({ template: 'supabase' })
// Use token to authenticate with external service
const response = await fetch('https://api.supabase.com/endpoint', {
headers: {
'Authorization': `Bearer ${token}`
}
})
}
```
**Cloudflare Workers**:
```typescript
import { clerkClient } from '@clerk/backend'
// Generate token for user
const token = await clerkClient.users.getUserOauthAccessToken(
userId,
'supabase' // Template name
)
```
---
## TypeScript Type Safety
Add type safety for custom claims with global declarations:
**Create `types/globals.d.ts`**:
```typescript
export {}
declare global {
interface CustomJwtSessionClaims {
metadata: {
role?: 'admin' | 'moderator' | 'user'
onboardingComplete?: boolean
department?: string
organizationId?: string
}
}
}
```
**Usage**:
```typescript
import { auth } from '@clerk/nextjs/server'
export default async function Page() {
const { sessionClaims } = await auth()
// TypeScript now knows about these properties
const role = sessionClaims?.metadata?.role // Type: 'admin' | 'moderator' | 'user' | undefined
const isComplete = sessionClaims?.metadata?.onboardingComplete // Type: boolean | undefined
}
```
---
## Common Use Cases
### 1. Role-Based Access Control (RBAC)
**JWT Template**:
```json
{
"email": "{{user.primary_email_address}}",
"role": "{{user.public_metadata.role || 'user'}}",
"permissions": "{{user.public_metadata.permissions}}"
}
```
**Backend Verification**:
```typescript
app.get('/api/admin', async (c) => {
const sessionClaims = c.get('sessionClaims')
if (sessionClaims?.role !== 'admin') {
return c.json({ error: 'Forbidden' }, 403)
}
return c.json({ message: 'Admin data' })
})
```
### 2. Multi-Tenant Applications
**JWT Template**:
```json
{
"user_id": "{{user.id}}",
"email": "{{user.primary_email_address}}",
"org_id": "{{user.public_metadata.org_id}}",
"org_slug": "{{user.public_metadata.org_slug}}",
"org_role": "{{user.public_metadata.org_role}}"
}
```
**Usage**:
```typescript
// Filter data by organization
const data = await db.query('SELECT * FROM items WHERE org_id = ?', [
sessionClaims.org_id
])
```
### 3. Supabase Integration
**JWT Template** (Name: `supabase`):
```json
{
"email": "{{user.primary_email_address}}",
"app_metadata": {
"provider": "clerk"
},
"user_metadata": {
"full_name": "{{user.full_name}}",
"avatar_url": "{{user.image_url}}"
}
}
```
**Usage**:
```typescript
import { createClient } from '@supabase/supabase-js'
const token = await getToken({ template: 'supabase' })
const supabase = createClient(
process.env.SUPABASE_URL,
process.env.SUPABASE_ANON_KEY,
{
global: {
headers: {
Authorization: `Bearer ${token}`
}
}
}
)
```
### 4. Grafbase GraphQL Integration
**JWT Template** (Name: `grafbase`):
```json
{
"sub": "{{user.id}}",
"groups": ["org:admin", "org:member"]
}
```
**Grafbase Configuration**:
```toml
[auth.providers.clerk]
issuer = "https://your-app.clerk.accounts.dev"
jwks = "https://your-app.clerk.accounts.dev/.well-known/jwks.json"
```
---
## Limitations & Gotchas
### 1. Token Size Limit: 1.2KB for Custom Claims
**Problem**: Browsers limit cookies to 4KB. Clerk's default claims consume ~2.8KB, leaving **1.2KB for custom claims**.
**Solution**:
- ✅ Use JWT for minimal claims (user ID, role, email)
- ✅ Store large data (bio, preferences) in your database
- ✅ Fetch additional data after authentication
- ❌ Don't include large objects or arrays
**Bad Example** (Exceeds limit):
```json
{
"email": "{{user.primary_email_address}}",
"bio": "{{user.public_metadata.bio}}", // 6KB bio field
"all_metadata": "{{user.public_metadata}}" // Entire object
}
```
**Good Example** (Under limit):
```json
{
"user_id": "{{user.id}}",
"email": "{{user.primary_email_address}}",
"role": "{{user.public_metadata.role}}"
}
```
### 2. Reserved Claims
These claims are **reserved** and cannot be used in custom templates:
| Claim | Reason |
|-------|--------|
| `azp` | Auto-included default claim |
| `exp` | Auto-included default claim |
| `iat` | Auto-included default claim |
| `iss` | Auto-included default claim |
| `jti` | Auto-included default claim |
| `nbf` | Auto-included default claim |
| `sub` | Auto-included default claim |
**Error**: Attempting to override reserved claims returns HTTP 400 with error code `jwt_template_reserved_claim`.
### 3. Session-Bound Claims
These claims are **session-specific** and cannot be included in **custom JWTs** (but ARE included in default session tokens):
| Claim | Description |
|-------|-------------|
| `sid` | Session ID |
| `v` | Version |
| `pla` | Plan |
| `fea` | Features |
**Why**: Custom JWTs are generated on-demand and not tied to a specific session.
**Solution**: If you need session data, use **custom session tokens** instead of custom JWTs.
### 4. Invalid Shortcodes Resolve to `null`
```json
{
"valid": "{{user.first_name}}",
"invalid": "{{user.i_dont_exist}}"
}
```
**Result**:
```json
{
"valid": "John",
"invalid": null
}
```
**Prevention**: Always test templates with real user data before deploying.
### 5. Custom JWTs May Incur Latency
- Default session tokens are cached
- Custom JWTs require fetching user data on-demand
- This can add **50-200ms** to token generation
**When to Use Custom JWTs**:
- ✅ Integrating with third-party services (Supabase, Hasura, Grafbase)
- ✅ API-to-API authentication
- ❌ Not recommended for every frontend request (use default session tokens)
### 6. Organization Data Limitations
- Only **active organization** data is included by default
- For all organizations, store in `user.public_metadata` and use custom template
- `orgs` claim was removed in Core 2 (breaking change)
### 7. Metadata Sync Timing
- Metadata changes may take a few seconds to propagate to tokens
- Use `user.reload()` in frontend to refresh user object
- Backend should always fetch fresh data for critical operations
---
## Official Documentation
- **JWT Templates Guide**: https://clerk.com/docs/guides/sessions/jwt-templates
- **Custom Session Tokens**: https://clerk.com/docs/backend-requests/making/custom-session-token
- **Session Claims Access**: https://clerk.com/docs/upgrade-guides/core-2/node
- **Backend SDK**: https://clerk.com/docs/reference/backend/overview
- **Supabase Integration**: https://clerk.com/docs/guides/development/integrations/databases/supabase
- **Grafbase Integration**: https://clerk.com/docs/guides/development/integrations/databases/grafbase
---
## Quick Reference: All Available Shortcodes
### User Properties
```
{{user.id}}
{{user.first_name}}
{{user.last_name}}
{{user.full_name}}
{{user.primary_email_address}}
{{user.primary_phone_address}}
{{user.image_url}}
{{user.created_at}}
{{user.username}}
{{user.email_verified}}
{{user.phone_number_verified}}
```
### Metadata
```
{{user.public_metadata}}
{{user.public_metadata.FIELD_NAME}}
{{user.public_metadata.nested.field}}
{{user.unsafe_metadata}}
{{user.unsafe_metadata.FIELD_NAME}}
```
### Operators
```
{{shortcode1 || shortcode2}} // Fallback
{{shortcode || 'default'}} // Default value
"{{shortcode1}} {{shortcode2}}" // String interpolation
```
---
**Last Updated**: 2025-10-22
**Verified Against**: Clerk API v2025-04-10
**Production Tested**: ✅ Multiple frameworks

575
references/testing-guide.md Normal file
View File

@@ -0,0 +1,575 @@
# Clerk Authentication - Testing Guide
**Last Updated**: 2025-10-28
**Source**: https://clerk.com/docs/guides/development/testing/overview
This guide covers testing Clerk authentication in your applications, including test credentials, session tokens, testing tokens, and E2E testing with Playwright.
---
## Overview
Testing authentication flows is essential for reliable applications. Clerk provides several tools to make testing easier:
1. **Test Emails & Phone Numbers** - Fake credentials with fixed OTP codes
2. **Session Tokens** - Generate valid tokens via Backend API
3. **Testing Tokens** - Bypass bot detection in test suites
4. **Framework Integrations** - Playwright and Cypress helpers
---
## Quick Start: Test Mode
### Enable Test Mode
Every **development instance** has test mode enabled by default.
For **production instances** (NOT recommended for real customer data):
1. Navigate to Clerk Dashboard → **Settings**
2. Enable the **Enable test mode** toggle
> **WARNING**: Never use test mode on instances managing actual customers.
---
## Test Emails & Phone Numbers
### Fake Email Addresses
Any email with the `+clerk_test` subaddress is a test email address:
```
jane+clerk_test@example.com
john+clerk_test@gmail.com
test+clerk_test@mycompany.com
```
**Behavior**:
- ✅ No emails sent (saves your email quota)
- ✅ Fixed OTP code: `424242`
- ✅ Works in all sign-up/sign-in flows
### Fake Phone Numbers
Any [fictional phone number](https://en.wikipedia.org/wiki/555_(telephone_number)) is a test phone number:
**Format**: `+1 (XXX) 555-0100` to `+1 (XXX) 555-0199`
**Examples**:
```
+12015550100
+19735550133
+14155550142
```
**Behavior**:
- ✅ No SMS sent (saves your SMS quota)
- ✅ Fixed OTP code: `424242`
- ✅ Works in all verification flows
### Monthly Limits (Development Instances)
Clerk development instances have limits:
- **20 SMS messages** per month
- **100 emails** per month
**Excluded from limits**:
- SMS to US numbers
- SMS/emails to test addresses (with `+clerk_test` or 555 numbers)
- Self-delivered messages (your own SMTP/SMS provider)
- Paid subscription apps
To request higher limits, contact Clerk support.
---
## Code Examples: Test Credentials
### Sign In with Test Email
```tsx
import { useSignIn } from '@clerk/clerk-react'
const testSignInWithEmailCode = async () => {
const { signIn } = useSignIn()
const emailAddress = 'john+clerk_test@example.com'
// Step 1: Create sign-in attempt
const signInResp = await signIn.create({ identifier: emailAddress })
// Step 2: Find email code factor
const { emailAddressId } = signInResp.supportedFirstFactors.find(
(ff) => ff.strategy === 'email_code' && ff.safeIdentifier === emailAddress,
)! as EmailCodeFactor
// Step 3: Prepare email verification
await signIn.prepareFirstFactor({
strategy: 'email_code',
emailAddressId: emailAddressId,
})
// Step 4: Verify with fixed code
const attemptResponse = await signIn.attemptFirstFactor({
strategy: 'email_code',
code: '424242', // Fixed test code
})
if (attemptResponse.status === 'complete') {
console.log('Sign in successful!')
} else {
console.error('Sign in failed')
}
}
```
### Sign Up with Test Phone
```tsx
import { useSignUp } from '@clerk/clerk-react'
const testSignUpWithPhoneNumber = async () => {
const { signUp } = useSignUp()
// Step 1: Create sign-up with test phone
await signUp.create({
phoneNumber: '+12015550100',
})
// Step 2: Prepare phone verification
await signUp.preparePhoneNumberVerification()
// Step 3: Verify with fixed code
const res = await signUp.attemptPhoneNumberVerification({
code: '424242', // Fixed test code
})
if (res.verifications.phoneNumber.status === 'verified') {
console.log('Sign up successful!')
} else {
console.error('Sign up failed')
}
}
```
---
## Session Tokens (Backend API)
For testing API endpoints or backend services, you need valid session tokens.
### Flow: Generate Session Token
**4-Step Process**:
1. **Create User** (if needed)
2. **Create Session** for user
3. **Create Session Token** from session ID
4. **Use Token** in Authorization header
### Step-by-Step Implementation
#### Step 1: Create User
**Endpoint**: `POST https://api.clerk.com/v1/users`
```bash
curl -X POST https://api.clerk.com/v1/users \
-H "Authorization: Bearer sk_test_YOUR_SECRET_KEY" \
-H "Content-Type: application/json" \
-d '{
"email_address": ["test+clerk_test@example.com"],
"password": "TestPassword123!"
}'
```
**Response**:
```json
{
"id": "user_2abc123def456",
"email_addresses": [
{
"id": "idn_2xyz789",
"email_address": "test+clerk_test@example.com"
}
]
}
```
**Save**: `user_id` for next step
#### Step 2: Create Session
**Endpoint**: `POST https://api.clerk.com/v1/sessions`
```bash
curl -X POST https://api.clerk.com/v1/sessions \
-H "Authorization: Bearer sk_test_YOUR_SECRET_KEY" \
-H "Content-Type: application/json" \
-d '{
"user_id": "user_2abc123def456"
}'
```
**Response**:
```json
{
"id": "sess_2xyz789abc123",
"user_id": "user_2abc123def456",
"status": "active"
}
```
**Save**: `session_id` for next step
#### Step 3: Create Session Token
**Endpoint**: `POST https://api.clerk.com/v1/sessions/{session_id}/tokens`
```bash
curl -X POST https://api.clerk.com/v1/sessions/sess_2xyz789abc123/tokens \
-H "Authorization: Bearer sk_test_YOUR_SECRET_KEY"
```
**Response**:
```json
{
"jwt": "eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9...",
"object": "token"
}
```
**Save**: `jwt` token
#### Step 4: Use Token in Requests
```bash
curl https://yourdomain.com/api/protected \
-H "Authorization: Bearer eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9..."
```
### Token Lifetime
**CRITICAL**: Session tokens are valid for **60 seconds only**.
**Refresh Strategies**:
**Option 1: Before Each Test**
```typescript
beforeEach(async () => {
sessionToken = await refreshSessionToken(sessionId)
})
```
**Option 2: Interval Timer**
```typescript
setInterval(async () => {
sessionToken = await refreshSessionToken(sessionId)
}, 50000) // Refresh every 50 seconds
```
### Node.js Script Example
See `scripts/generate-session-token.js` for a complete implementation.
---
## Testing Tokens (Bot Detection Bypass)
Testing Tokens bypass Clerk's bot detection mechanisms during automated testing.
### What Are Testing Tokens?
- **Purpose**: Prevent "Bot traffic detected" errors in test suites
- **Lifetime**: Short-lived (expires after use)
- **Scope**: Valid only for specific Clerk instance
- **Source**: Obtained via Backend API
### When to Use
**Use Testing Tokens when**:
- Running E2E tests with Playwright or Cypress
- Automated test suites triggering bot detection
- CI/CD pipelines running authentication flows
**Alternatives**:
- Use `@clerk/testing` package (handles automatically)
- Playwright integration (recommended)
- Cypress integration (recommended)
### Obtain Testing Token
**Endpoint**: `POST https://api.clerk.com/v1/testing_tokens`
```bash
curl -X POST https://api.clerk.com/v1/testing_tokens \
-H "Authorization: Bearer sk_test_YOUR_SECRET_KEY"
```
**Response**:
```json
{
"token": "1713877200-c_2J2MvPu9PnXcuhbPZNao0LOXqK9A7YrnBn0HmIWxy"
}
```
### Use Testing Token
Add `__clerk_testing_token` query parameter to Frontend API requests:
```
POST https://happy-hippo-1.clerk.accounts.dev/v1/client/sign_ups?__clerk_testing_token=1713877200-c_2J2MvPu9PnXcuhbPZNao0LOXqK9A7YrnBn0HmIWxy
```
### Production Limitations
Testing Tokens work in **both development and production**, but:
**❌ Not Supported in Production**:
- Code-based authentication (SMS OTP, Email OTP)
**✅ Supported in Production**:
- Email + password authentication
- Email magic link (sign-in via email)
---
## E2E Testing with Playwright
Clerk provides first-class Playwright support via `@clerk/testing`.
### Install
```bash
npm install -D @clerk/testing
```
### Set Environment Variables
In your test runner (e.g., `.env.test` or GitHub Actions secrets):
```bash
CLERK_PUBLISHABLE_KEY=pk_test_...
CLERK_SECRET_KEY=sk_test_...
```
**Security**: Use GitHub Actions secrets or similar for CI/CD.
### Global Setup
Configure `clerkSetup()` to obtain Testing Token once for all tests:
**File**: `global.setup.ts`
```typescript
import { clerkSetup } from '@clerk/testing/playwright'
import { test as setup } from '@playwright/test'
// Run setup serially (important for fully parallel Playwright config)
setup.describe.configure({ mode: 'serial' })
setup('global setup', async ({}) => {
await clerkSetup()
})
```
**What This Does**:
- Obtains Testing Token from Clerk Backend API
- Stores token in `CLERK_TESTING_TOKEN` environment variable
- Makes token available for all tests
### Use in Tests
Import `setupClerkTestingToken()` and call before navigating to auth pages:
**File**: `auth.spec.ts`
```typescript
import { setupClerkTestingToken } from '@clerk/testing/playwright'
import { test, expect } from '@playwright/test'
test('sign up flow', async ({ page }) => {
// Inject Testing Token for this test
await setupClerkTestingToken({ page })
// Navigate to sign-up page
await page.goto('/sign-up')
// Fill form with test credentials
await page.fill('input[name="emailAddress"]', 'test+clerk_test@example.com')
await page.fill('input[name="password"]', 'TestPassword123!')
await page.click('button[type="submit"]')
// Verify with fixed OTP
await page.fill('input[name="code"]', '424242')
await page.click('button[type="submit"]')
// Assert success
await expect(page).toHaveURL('/dashboard')
})
test('sign in with test phone', async ({ page }) => {
await setupClerkTestingToken({ page })
await page.goto('/sign-in')
await page.fill('input[name="identifier"]', '+12015550100')
await page.click('button[type="submit"]')
// Enter fixed OTP
await page.fill('input[name="code"]', '424242')
await page.click('button[type="submit"]')
await expect(page).toHaveURL('/dashboard')
})
```
### Manual Testing Token Setup (Alternative)
Instead of `clerkSetup()`, manually set the environment variable:
```bash
export CLERK_TESTING_TOKEN="1713877200-c_2J2MvPu9PnXcuhbPZNao0LOXqK9A7YrnBn0HmIWxy"
```
Then run tests as usual. `setupClerkTestingToken()` will use this value.
### Demo Repository
Clerk provides a complete example:
**Repository**: https://github.com/clerk/clerk-playwright-nextjs
**Features**:
- Next.js App Router with Clerk
- Playwright E2E tests
- Testing Tokens setup
- Test user authentication
**To Run**:
1. Clone repo
2. Add Clerk API keys to `.env.local`
3. Create test user with username + password
4. Enable username/password auth in Clerk Dashboard
5. Run `npm test`
---
## Testing Email Links (Magic Links)
Email links are challenging to test in E2E suites.
**Recommendation**: Use email verification codes instead.
### Enable Email Verification Code
1. Clerk Dashboard → **Email, Phone, Username**
2. Enable **Email verification code** strategy
3. Use the code flow in tests (easier to automate)
**Code flow** and **link flow** are functionally equivalent for most use cases.
---
## Best Practices
### Development Testing
**Do**:
- Use test emails (`+clerk_test`) and phone numbers (555-01XX)
- Fixed OTP: `424242`
- Enable test mode in Clerk Dashboard
- Use `@clerk/testing` for Playwright/Cypress
**Don't**:
- Send real emails/SMS during tests (wastes quota)
- Use production keys in tests
- Enable test mode on production instances with real users
### Backend/API Testing
**Do**:
- Generate session tokens via Backend API
- Refresh tokens before each test or on interval
- Use test users created via API
- Store `CLERK_SECRET_KEY` securely
**Don't**:
- Hardcode session tokens (expire in 60 seconds)
- Reuse expired tokens
- Expose secret keys in logs or version control
### E2E Testing
**Do**:
- Use `@clerk/testing` for automatic Testing Token management
- Configure global setup for token generation
- Use test credentials in all flows
- Run tests in CI/CD with secret environment variables
**Don't**:
- Skip `setupClerkTestingToken()` (triggers bot detection)
- Manually implement Testing Token logic (use helpers)
- Test code-based auth in production with Testing Tokens
---
## Troubleshooting
### "Bot traffic detected" Error
**Cause**: Missing Testing Token in test suite
**Solution**:
1. Install `@clerk/testing`
2. Configure `clerkSetup()` in global setup
3. Call `setupClerkTestingToken({ page })` in tests
### Session Token Expired
**Cause**: Token lifetime is 60 seconds
**Solution**:
- Refresh token before each test: `beforeEach(() => refreshToken())`
- Use interval timer: `setInterval(() => refreshToken(), 50000)`
### Test Email Not Working
**Cause**: Test mode not enabled, or wrong email format
**Solution**:
- Ensure email has `+clerk_test` subaddress
- Enable test mode in Clerk Dashboard
- Use fixed OTP: `424242`
### 20 SMS / 100 Email Limit Reached
**Cause**: Exceeded monthly limit for development instance
**Solution**:
- Use test credentials (excluded from limits)
- Contact Clerk support to request higher limits
- Use self-delivered SMS/email provider
---
## Reference Links
**Official Docs**:
- Testing Overview: https://clerk.com/docs/guides/development/testing/overview
- Test Emails & Phones: https://clerk.com/docs/guides/development/testing/test-emails-and-phones
- Playwright Integration: https://clerk.com/docs/guides/development/testing/playwright/overview
- Backend API Reference: https://clerk.com/docs/reference/backend-api
**Packages**:
- `@clerk/testing`: https://github.com/clerk/javascript/tree/main/packages/testing
- Demo Repository: https://github.com/clerk/clerk-playwright-nextjs
**API Endpoints**:
- Create User: `POST /v1/users`
- Create Session: `POST /v1/sessions`
- Create Session Token: `POST /v1/sessions/{session_id}/tokens`
- Create Testing Token: `POST /v1/testing_tokens`
---
**Last Updated**: 2025-10-28