opencodekit 0.15.4 → 0.15.5

package/dist/template/.opencode/skill/cloudflare/references/ai-gateway/README.md

@@ -0,0 +1,695 @@
+# Cloudflare AI Gateway Skill
+
+Expert guidance for implementing and configuring Cloudflare AI Gateway - a universal gateway for AI model providers with analytics, caching, rate limiting, and routing capabilities.
+
+## When to Use This Skill
+
+- Setting up AI Gateway for any AI provider (OpenAI, Anthropic, Workers AI, etc.)
+- Implementing caching, rate limiting, or request retry/fallback
+- Configuring dynamic routing with A/B testing or model fallbacks
+- Managing provider API keys securely with BYOK
+- Setting up observability with logging and custom metadata
+- Integrating AI Gateway with Cloudflare Workers or external applications
+- Debugging AI Gateway requests or optimizing configurations
+
+## Core Concepts
+
+### Gateway Architecture
+
+AI Gateway acts as a proxy between your application and AI providers:
+
+```
+Your App → AI Gateway → AI Provider (OpenAI, Anthropic, etc.)
+         ↓
+    Analytics, Caching, Rate Limiting, Logging
+```
+
+**Key URL patterns:**
+- Unified API (OpenAI-compatible): `https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/compat/chat/completions`
+- Provider-specific: `https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/{provider}/{endpoint}`
+- Dynamic routes: Use route name instead of model: `dynamic/{route-name}`
+
+### Gateway Types
+
+1. **Unauthenticated Gateway**: Open access (not recommended for production)
+2. **Authenticated Gateway**: Requires `cf-aig-authorization` header with Cloudflare API token (recommended)
+
+### Provider Authentication Options
+
+1. **Unified Billing**: Use AI Gateway billing to pay for inference
+2. **BYOK (Store Keys)**: Store provider API keys in Cloudflare dashboard
+3. **Request Headers**: Include provider API key in each request
+
+## Common Patterns
+
+### Pattern 1: OpenAI SDK with Unified API Endpoint
+
+Most common pattern - drop-in replacement for OpenAI API with multi-provider support.
+
+```typescript
+import OpenAI from 'openai';
+
+const client = new OpenAI({
+  apiKey: process.env.OPENAI_API_KEY, // or any provider's key
+  baseURL: `https://gateway.ai.cloudflare.com/v1/${accountId}/${gatewayId}/compat`,
+  defaultHeaders: {
+    // Only needed for authenticated gateways
+    'cf-aig-authorization': `Bearer ${cfToken}`
+  }
+});
+
+// Switch providers by changing model format: {provider}/{model}
+const response = await client.chat.completions.create({
+  model: 'openai/gpt-4o-mini', // or 'anthropic/claude-sonnet-4-5'
+  messages: [{ role: 'user', content: 'Hello!' }]
+});
+```
+
+**Benefits:**
+- Works with existing OpenAI SDK tooling
+- Switch providers without code changes (just change model param)
+- Compatible with most OpenAI-compatible tools
+
+### Pattern 2: Provider-Specific Endpoints
+
+Use when you need the original provider's API schema.
+
+```typescript
+import OpenAI from 'openai';
+
+const client = new OpenAI({
+  apiKey: process.env.OPENAI_API_KEY,
+  baseURL: `https://gateway.ai.cloudflare.com/v1/${accountId}/${gatewayId}/openai`
+});
+
+// Standard OpenAI request - AI Gateway features still apply
+const response = await client.chat.completions.create({
+  model: 'gpt-4o-mini',
+  messages: [{ role: 'user', content: 'Hello!' }]
+});
+```
+
+### Pattern 3: Workers AI Binding with Gateway
+
+For Cloudflare Workers using Workers AI.
+
+```typescript
+export default {
+  async fetch(request, env, ctx) {
+    const response = await env.AI.run(
+      '@cf/meta/llama-3-8b-instruct',
+      { 
+        messages: [{ role: 'user', content: 'Hello!' }]
+      },
+      { 
+        gateway: { 
+          id: 'my-gateway',
+          metadata: { userId: '123', team: 'engineering' }
+        } 
+      }
+    );
+    
+    return new Response(JSON.stringify(response));
+  }
+};
+```
+
+### Pattern 4: Custom Metadata for Tracking
+
+Tag requests with user IDs, teams, or other identifiers (max 5 metadata entries).
+
+```typescript
+const response = await openai.chat.completions.create(
+  {
+    model: 'gpt-4o-mini',
+    messages: [{ role: 'user', content: 'Hello!' }]
+  },
+  {
+    headers: {
+      'cf-aig-metadata': JSON.stringify({
+        userId: 'user123',
+        team: 'engineering',
+        environment: 'production',
+        requestType: 'chat',
+        internal: true
+      })
+    }
+  }
+);
+```
+
+### Pattern 5: Per-Request Caching Control
+
+Override default gateway caching settings per request.
+
+```bash
+# Skip cache for this request
+curl https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/openai/chat/completions \
+  --header 'Authorization: Bearer $TOKEN' \
+  --header 'cf-aig-skip-cache: true' \
+  --data '{"model": "gpt-4o-mini", "messages": [...]}'
+
+# Custom cache TTL (1 hour)
+curl https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/openai/chat/completions \
+  --header 'Authorization: Bearer $TOKEN' \
+  --header 'cf-aig-cache-ttl: 3600' \
+  --data '{"model": "gpt-4o-mini", "messages": [...]}'
+
+# Custom cache key for deterministic caching
+curl https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/openai/chat/completions \
+  --header 'Authorization: Bearer $TOKEN' \
+  --header 'cf-aig-cache-key: greeting-response' \
+  --data '{"model": "gpt-4o-mini", "messages": [...]}'
+```
+
+**Cache headers:**
+- `cf-aig-skip-cache: true` - Bypass cache
+- `cf-aig-cache-ttl: <seconds>` - Custom TTL (min: 60s, max: 1 month)
+- `cf-aig-cache-key: <key>` - Custom cache key
+- Response header `cf-aig-cache-status: HIT|MISS` indicates cache status
+
+### Pattern 6: BYOK (Bring Your Own Keys)
+
+Store provider keys in dashboard, remove from code.
+
+**Setup:**
+1. Enable authentication on gateway
+2. Dashboard → AI Gateway → Select gateway → Provider Keys → Add API Key
+3. Remove provider API keys from code:
+
+```typescript
+// Before BYOK: Include provider key in every request
+const client = new OpenAI({
+  apiKey: process.env.OPENAI_API_KEY, // Provider key
+  baseURL: `https://gateway.ai.cloudflare.com/v1/${accountId}/${gatewayId}/openai`,
+  defaultHeaders: {
+    'cf-aig-authorization': `Bearer ${cfToken}` // Gateway auth
+  }
+});
+
+// After BYOK: Only gateway auth needed
+const client = new OpenAI({
+  // No apiKey needed - stored in dashboard
+  baseURL: `https://gateway.ai.cloudflare.com/v1/${accountId}/${gatewayId}/openai`,
+  defaultHeaders: {
+    'cf-aig-authorization': `Bearer ${cfToken}` // Only gateway auth
+  }
+});
+```
+
+### Pattern 7: Dynamic Routing with Fallbacks
+
+Configure routing logic in dashboard, not code.
+
+```typescript
+// Use route name instead of model
+const response = await client.chat.completions.create({
+  model: 'dynamic/support', // Route name from dashboard
+  messages: [{ role: 'user', content: 'Hello!' }]
+});
+```
+
+**Dynamic routing use cases:**
+- A/B testing between models
+- Rate/budget limits per user/team
+- Model fallbacks on errors
+- Conditional routing (paid vs free users)
+
+**Route configuration (in dashboard):**
+1. Create route: Dashboard → Gateway → Dynamic Routes → Add Route
+2. Define flow with nodes:
+   - **Conditional**: Branch on metadata (e.g., `user.plan == "paid"`)
+   - **Percentage**: A/B split (e.g., 80% model A, 20% model B)
+   - **Rate Limit**: Quota enforcement, fallback when exceeded
+   - **Budget Limit**: Cost quota enforcement
+   - **Model**: Call specific provider/model
+3. Save & deploy version
+
+### Pattern 8: Error Handling
+
+```typescript
+try {
+  const response = await client.chat.completions.create({
+    model: 'gpt-4o-mini',
+    messages: [{ role: 'user', content: 'Hello!' }]
+  });
+} catch (error) {
+  // Rate limit exceeded
+  if (error.status === 429) {
+    console.error('Rate limit exceeded:', error.message);
+    // Implement backoff or use dynamic routing with fallback
+  }
+  
+  // Gateway authentication failed
+  if (error.status === 401) {
+    console.error('Gateway authentication failed - check cf-aig-authorization token');
+  }
+  
+  // Provider authentication failed
+  if (error.status === 403) {
+    console.error('Provider authentication failed - check API key or BYOK setup');
+  }
+  
+  throw error;
+}
+```
+
+## Configuration Reference
+
+### Dashboard Setup
+
+**Create gateway:**
+```bash
+# Via Dashboard: AI > AI Gateway > Create Gateway
+# Or via API:
+curl https://api.cloudflare.com/client/v4/accounts/{account_id}/ai-gateway/gateways \
+  -H "Authorization: Bearer $CF_API_TOKEN" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "id": "my-gateway",
+    "cache_ttl": 3600,
+    "cache_invalidate_on_update": true,
+    "rate_limiting_interval": 60,
+    "rate_limiting_limit": 100,
+    "rate_limiting_technique": "sliding",
+    "collect_logs": true
+  }'
+```
+
+### Feature Configuration
+
+**Caching:**
+- Dashboard: Settings → Cache Responses → Enable
+- Default TTL: Set in gateway settings
+- Cache behavior: Only for identical requests (text & image responses)
+- Use case: Support bots with limited prompt options
+
+**Rate Limiting:**
+- Dashboard: Settings → Rate-limiting → Enable
+- Parameters:
+  - Limit: Number of requests
+  - Interval: Time period (seconds)
+  - Technique: `fixed` or `sliding` window
+- Response: `429 Too Many Requests` when exceeded
+
+**Logging:**
+- Dashboard: Settings → Logs
+- Default: Enabled (up to 10M logs per gateway)
+- Per-request: `cf-aig-collect-log: false` to skip
+- Auto-delete: Enable to remove oldest logs when limit reached
+- Filter logs by: status, cache, provider, model, cost, tokens, duration, metadata
+
+### Wrangler Integration
+
+**Gateway with Workers AI:**
+
+```toml
+# wrangler.toml
+name = "my-worker"
+main = "src/index.ts"
+compatibility_date = "2024-01-01"
+
+[ai]
+binding = "AI"
+
+[[ai.gateway]]
+id = "my-gateway"
+```
+
+```typescript
+// src/index.ts
+export default {
+  async fetch(request, env, ctx): Promise<Response> {
+    const response = await env.AI.run(
+      '@cf/meta/llama-3-8b-instruct',
+      { prompt: 'Hello!' },
+      { gateway: { id: 'my-gateway' } }
+    );
+    
+    return Response.json(response);
+  }
+} satisfies ExportedHandler<Env>;
+```
+
+**Environment variables for gateways:**
+
+```toml
+# wrangler.toml
+[vars]
+CF_ACCOUNT_ID = "your-account-id"
+GATEWAY_ID = "my-gateway"
+
+# Secrets (use wrangler secret put)
+# CF_API_TOKEN - for authenticated gateways
+# OPENAI_API_KEY - if not using BYOK
+```
+
+```bash
+# Set secrets
+wrangler secret put CF_API_TOKEN
+wrangler secret put OPENAI_API_KEY
+```
+
+### API Token Permissions
+
+**For gateway management:**
+- AI Gateway - Read
+- AI Gateway - Edit
+
+**For authenticated gateway access:**
+- Create API token with appropriate permissions
+- Pass in `cf-aig-authorization: Bearer {token}` header
+
+## Supported Providers
+
+AI Gateway works with 15+ providers via unified API or provider-specific endpoints:
+
+| Provider | Unified API | Provider Endpoint | Notes |
+|----------|-------------|-------------------|-------|
+| OpenAI | ✅ `openai/gpt-4o` | `/openai/*` | Full support |
+| Anthropic | ✅ `anthropic/claude-3-5-sonnet` | `/anthropic/*` | Full support |
+| Google AI Studio | ✅ `google-ai-studio/gemini-2.0-flash` | `/google-ai-studio/*` | Full support |
+| Workers AI | ✅ `workersai/@cf/meta/llama-3` | `/workers-ai/*` | Native integration |
+| Azure OpenAI | ✅ `azure-openai/*` | `/azure-openai/*` | Deployment names |
+| AWS Bedrock | ❌ | `/bedrock/*` | Provider endpoint only |
+| Groq | ✅ `groq/*` | `/groq/*` | Fast inference |
+| Mistral | ✅ `mistral/*` | `/mistral/*` | Full support |
+| Cohere | ✅ `cohere/*` | `/cohere/*` | Full support |
+| Perplexity | ✅ `perplexity/*` | `/perplexity/*` | Full support |
+| xAI (Grok) | ✅ `grok/*` | `/grok/*` | Full support |
+| DeepSeek | ✅ `deepseek/*` | `/deepseek/*` | Full support |
+| Cerebras | ✅ `cerebras/*` | `/cerebras/*` | Fast inference |
+| Replicate | ❌ | `/replicate/*` | Provider endpoint only |
+| HuggingFace | ❌ | `/huggingface/*` | Provider endpoint only |
+
+See [full provider list](https://developers.cloudflare.com/ai-gateway/usage/providers/)
+
+## Observability
+
+### Analytics Dashboard
+
+View in Dashboard → AI Gateway → Select gateway:
+- Request count over time
+- Token usage (input/output)
+- Cost tracking (estimated or custom)
+- Cache hit rate
+- Error rates by provider/model
+- Latency percentiles
+
+### Log Structure
+
+Each log entry contains:
+- User prompt & model response
+- Provider & model
+- Timestamp
+- Request status (success/error)
+- Token usage (input/output/total)
+- Cost
+- Duration (ms)
+- Cache status (HIT/MISS)
+- Custom metadata
+- Request/Event ID
+
+### Custom Cost Tracking
+
+For custom models or providers not in Cloudflare's pricing database:
+
+```bash
+# Dashboard: Gateway → Settings → Custom Costs
+# Or via API:
+curl https://api.cloudflare.com/client/v4/accounts/{account_id}/ai-gateway/gateways/{gateway_id}/custom-costs \
+  -H "Authorization: Bearer $CF_API_TOKEN" \
+  -d '{
+    "model": "custom-model-v1",
+    "input_cost": 0.01,
+    "output_cost": 0.03
+  }'
+```
+
+## Advanced Use Cases
+
+### Multi-Model Chat with Fallbacks
+
+```typescript
+// Configure in dashboard:
+// Route: dynamic/smart-chat
+// - Try GPT-4 first
+// - Fallback to Claude if error
+// - Fallback to Llama if both fail
+
+const response = await client.chat.completions.create({
+  model: 'dynamic/smart-chat',
+  messages: [{ role: 'user', content: 'Complex reasoning task' }]
+});
+```
+
+### A/B Testing Models
+
+```typescript
+// Dashboard: Create route with Percentage node
+// - 50% to gpt-4o-mini
+// - 50% to claude-sonnet-4-5
+// Analyze logs to compare quality/cost/latency
+
+const response = await client.chat.completions.create({
+  model: 'dynamic/ab-test',
+  messages: [{ role: 'user', content: prompt }],
+  // Add metadata to track experiments
+  headers: {
+    'cf-aig-metadata': JSON.stringify({ experiment: 'model-comparison-v1' })
+  }
+});
+```
+
+### User-Based Rate Limiting
+
+```typescript
+// Dashboard: Create route with Rate Limit node
+// - Condition: Check metadata.userId
+// - Limit: 100 requests/hour per user
+// - Fallback: Return error or use cheaper model
+
+const response = await client.chat.completions.create(
+  {
+    model: 'dynamic/user-limited',
+    messages: [{ role: 'user', content: prompt }]
+  },
+  {
+    headers: {
+      'cf-aig-metadata': JSON.stringify({ userId })
+    }
+  }
+);
+```
+
+### Semantic Caching (Future)
+
+Currently, caching requires identical requests. Semantic caching (similar but not identical requests) is planned.
+
+**Current workaround:**
+```typescript
+// Use cf-aig-cache-key for grouped responses
+const normalizedPrompt = normalizePrompt(userInput); // Your logic
+const cacheKey = hashPrompt(normalizedPrompt);
+
+const response = await fetch(gatewayUrl, {
+  headers: {
+    'cf-aig-cache-key': cacheKey,
+    'cf-aig-cache-ttl': '3600'
+  },
+  // ... rest of request
+});
+```
+
+## Debugging & Troubleshooting
+
+### Check Gateway Status
+
+```bash
+# List all gateways
+curl https://api.cloudflare.com/client/v4/accounts/{account_id}/ai-gateway/gateways \
+  -H "Authorization: Bearer $CF_API_TOKEN"
+
+# Get specific gateway
+curl https://api.cloudflare.com/client/v4/accounts/{account_id}/ai-gateway/gateways/{gateway_id} \
+  -H "Authorization: Bearer $CF_API_TOKEN"
+```
+
+### Inspect Request Logs
+
+Dashboard → Gateway → Logs
+
+**Filter examples:**
+- `status: error` - All failed requests
+- `provider: openai` - OpenAI requests only
+- `metadata.userId: user123` - Specific user
+- `cache: not cached` - Cache misses only
+- `cost > 0.01` - Expensive requests
+
+### Common Issues
+
+**401 Unauthorized:**
+- Authenticated gateway without `cf-aig-authorization` header
+- Invalid/expired CF API token
+- Check token permissions (AI Gateway - Read)
+
+**403 Forbidden:**
+- Provider API key invalid/missing
+- BYOK key not configured or expired
+- Provider quota exceeded
+
+**429 Rate Limited:**
+- Gateway rate limit exceeded
+- Check settings: Dashboard → Gateway → Settings → Rate-limiting
+- Implement backoff or use dynamic routing
+
+**Cache not working:**
+- Requests must be identical (body, model, parameters)
+- Caching only supports text/image responses
+- Check `cf-aig-cache-status` header in response
+- Verify caching enabled: Dashboard → Settings → Cache Responses
+
+**Logs not appearing:**
+- Check log limit (default: 10M per gateway)
+- Verify logs enabled: Dashboard → Settings → Logs
+- Per-request `cf-aig-collect-log: false` bypasses logging
+- Wait 30-60s for logs to appear
+
+## API Reference
+
+### Gateway Management
+
+```bash
+# Create gateway
+POST /accounts/{account_id}/ai-gateway/gateways
+
+# Update gateway
+PUT /accounts/{account_id}/ai-gateway/gateways/{gateway_id}
+
+# Delete gateway
+DELETE /accounts/{account_id}/ai-gateway/gateways/{gateway_id}
+
+# List gateways
+GET /accounts/{account_id}/ai-gateway/gateways
+```
+
+### Log Management
+
+```bash
+# Get logs
+GET /accounts/{account_id}/ai-gateway/gateways/{gateway_id}/logs
+
+# Delete logs
+DELETE /accounts/{account_id}/ai-gateway/gateways/{gateway_id}/logs
+
+# Filter logs (query params)
+?status=error&provider=openai&cache=not_cached
+```
+
+### Headers Reference
+
+**Gateway authentication:**
+- `cf-aig-authorization: Bearer {token}` - Required for authenticated gateways
+
+**Caching:**
+- `cf-aig-cache-ttl: {seconds}` - Cache duration (60s - 1 month)
+- `cf-aig-skip-cache: true` - Bypass cache
+- `cf-aig-cache-key: {key}` - Custom cache key
+- Response: `cf-aig-cache-status: HIT|MISS`
+
+**Logging:**
+- `cf-aig-collect-log: false` - Skip logging for this request
+
+**Metadata:**
+- `cf-aig-metadata: {json}` - Custom tracking data (max 5 entries, string/number/boolean values)
+
+## Best Practices
+
+1. **Always use authenticated gateways in production**
+   - Prevents unauthorized access
+   - Protects against log storage abuse
+   - Required for BYOK
+
+2. **Use BYOK for provider keys**
+   - Removes keys from codebase
+   - Easier key rotation
+   - Centralized management
+
+3. **Add custom metadata to all requests**
+   - Track users, teams, environments
+   - Filter logs effectively
+   - Debug production issues faster
+
+4. **Configure appropriate rate limits**
+   - Prevent runaway costs
+   - Use dynamic routing for per-user limits
+   - Combine with budget limits
+
+5. **Enable caching for deterministic prompts**
+   - Support bots with fixed options
+   - Static content generation
+   - Reduces costs & latency
+
+6. **Use dynamic routing for resilience**
+   - Model fallbacks on errors
+   - A/B testing without code changes
+   - Gradual rollouts
+
+7. **Monitor logs regularly**
+   - Set up automatic log deletion
+   - Export logs for long-term analysis
+   - Track cost trends
+
+8. **Test with provider-specific endpoints first**
+   - Validates provider integration
+   - Easier debugging
+   - Migrate to unified API after validation
+
+## Examples Repository
+
+See real-world usage:
+- [NextChat](https://github.com/ChatGPTNextWeb/NextChat/blob/main/app/utils/cloudflare.ts) - URL parsing utilities
+- [LibreChat](https://github.com/danny-avila/LibreChat) - Multi-provider chat with AI Gateway
+- [Continue.dev](https://github.com/continuedev/continue/blob/main/core/llm/llms/Cloudflare.ts) - IDE integration
+- [Big-AGI](https://github.com/enricoros/big-AGI) - Complex gateway path handling
+
+## Resources
+
+- [Official Docs](https://developers.cloudflare.com/ai-gateway/)
+- [API Reference](https://developers.cloudflare.com/api/resources/ai_gateway/)
+- [Provider Guides](https://developers.cloudflare.com/ai-gateway/usage/providers/)
+- [Workers AI Integration](https://developers.cloudflare.com/workers-ai/)
+- [Discord Community](https://discord.cloudflare.com)
+
+## Quick Reference
+
+**Create gateway:**
+```bash
+Dashboard → AI → AI Gateway → Create Gateway
+```
+
+**Basic request:**
+```typescript
+const client = new OpenAI({
+  baseURL: `https://gateway.ai.cloudflare.com/v1/${accountId}/${gatewayId}/compat`
+});
+```
+
+**Check cache status:**
+```bash
+# Response header: cf-aig-cache-status: HIT|MISS
+```
+
+**Get account/gateway IDs:**
+```bash
+# Account ID: Dashboard → Overview → Account ID
+# Gateway ID: Dashboard → AI Gateway → Gateway name/ID
+```
+
+**Required env vars:**
+```bash
+CF_ACCOUNT_ID=xxx
+GATEWAY_ID=xxx
+CF_API_TOKEN=xxx  # For authenticated gateways
+PROVIDER_API_KEY=xxx  # If not using BYOK
+```

package/dist/template/.opencode/skill/cloudflare/references/ai-search/README.md

@@ -0,0 +1,14 @@
+# Cloudflare AI Search Skill Reference
+
+Expert guidance for implementing Cloudflare AI Search (formerly AutoRAG), Cloudflare's managed semantic search and RAG service....
+
+## In This Reference
+
+- **[configuration.md](./configuration.md)** - Setup, deployment, configuration
+- **[api.md](./api.md)** - API endpoints, methods, interfaces
+- **[patterns.md](./patterns.md)** - Common patterns, use cases, examples
+- **[gotchas.md](./gotchas.md)** - Troubleshooting, best practices, limitations
+
+## See Also
+
+- [Cloudflare Docs](https://developers.cloudflare.com/)