perplexity-known-pitfallsClaude Skill
Identify and avoid Perplexity anti-patterns and common integration mistakes.
| name | perplexity-known-pitfalls |
| description | Identify and avoid Perplexity anti-patterns and common integration mistakes. Use when reviewing Perplexity code for issues, onboarding new developers, or auditing existing Perplexity integrations for best practices violations. Trigger with phrases like "perplexity mistakes", "perplexity anti-patterns", "perplexity pitfalls", "perplexity what not to do", "perplexity code review". |
| allowed-tools | Read, Grep |
| version | 1.0.0 |
| license | MIT |
| author | Jeremy Longshore <jeremy@intentsolutions.io> |
Perplexity Known Pitfalls
Overview
Common mistakes and anti-patterns when integrating with Perplexity.
Prerequisites
- Access to Perplexity codebase for review
- Understanding of async/await patterns
- Knowledge of security best practices
- Familiarity with rate limiting concepts
Pitfall #1: Synchronous API Calls in Request Path
❌ Anti-Pattern
// User waits for Perplexity API call app.post('/checkout', async (req, res) => { const payment = await perplexityClient.processPayment(req.body); // 2-5s latency const notification = await perplexityClient.sendEmail(payment); // Another 1-2s res.json({ success: true }); // User waited 3-7s });
✅ Better Approach
// Return immediately, process async app.post('/checkout', async (req, res) => { const jobId = await queue.enqueue('process-checkout', req.body); res.json({ jobId, status: 'processing' }); // 50ms response }); // Background job async function processCheckout(data) { const payment = await perplexityClient.processPayment(data); await perplexityClient.sendEmail(payment); }
Pitfall #2: Not Handling Rate Limits
❌ Anti-Pattern
// Blast requests, crash on 429 for (const item of items) { await perplexityClient.process(item); // Will hit rate limit }
✅ Better Approach
import pLimit from 'p-limit'; const limit = pLimit(5); // Max 5 concurrent const rateLimiter = new RateLimiter({ tokensPerSecond: 10 }); for (const item of items) { await rateLimiter.acquire(); await limit(() => perplexityClient.process(item)); }
Pitfall #3: Leaking API Keys
❌ Anti-Pattern
// In frontend code (visible to users!) const client = new PerplexityClient({ apiKey: 'sk_live_ACTUAL_KEY_HERE', // Anyone can see this }); // In git history git commit -m "add API key" // Exposed forever
✅ Better Approach
// Backend only, environment variable const client = new PerplexityClient({ apiKey: process.env.PERPLEXITY_API_KEY, }); // Use .gitignore .env .env.local .env.*.local
Pitfall #4: Ignoring Idempotency
❌ Anti-Pattern
// Network error on response = duplicate charge! try { await perplexityClient.charge(order); } catch (error) { if (error.code === 'NETWORK_ERROR') { await perplexityClient.charge(order); // Charged twice! } }
✅ Better Approach
const idempotencyKey = `order-${order.id}-${Date.now()}`; await perplexityClient.charge(order, { idempotencyKey, // Safe to retry });
Pitfall #5: Not Validating Webhooks
❌ Anti-Pattern
// Trust any incoming request app.post('/webhook', (req, res) => { processWebhook(req.body); // Attacker can send fake events res.sendStatus(200); });
✅ Better Approach
app.post('/webhook', express.raw({ type: 'application/json' }), (req, res) => { const signature = req.headers['x-perplexity-signature']; if (!verifyPerplexitySignature(req.body, signature)) { return res.sendStatus(401); } processWebhook(JSON.parse(req.body)); res.sendStatus(200); } );
Pitfall #6: Missing Error Handling
❌ Anti-Pattern
// Crashes on any error const result = await perplexityClient.get(id); console.log(result.data.nested.value); // TypeError if missing
✅ Better Approach
try { const result = await perplexityClient.get(id); console.log(result?.data?.nested?.value ?? 'default'); } catch (error) { if (error instanceof PerplexityNotFoundError) { return null; } if (error instanceof PerplexityRateLimitError) { await sleep(error.retryAfter); return this.get(id); // Retry } throw error; // Rethrow unknown errors }
Pitfall #7: Hardcoding Configuration
❌ Anti-Pattern
const client = new PerplexityClient({ timeout: 5000, // Too short for some operations baseUrl: 'https://api.perplexity.com', // Can't change for staging });
✅ Better Approach
const client = new PerplexityClient({ timeout: parseInt(process.env.PERPLEXITY_TIMEOUT || '30000'), baseUrl: process.env.PERPLEXITY_BASE_URL || 'https://api.perplexity.com', });
Pitfall #8: Not Implementing Circuit Breaker
❌ Anti-Pattern
// When Perplexity is down, every request hangs for (const user of users) { await perplexityClient.sync(user); // All timeout sequentially }
✅ Better Approach
import CircuitBreaker from 'opossum'; const breaker = new CircuitBreaker(perplexityClient.sync, { timeout: 10000, errorThresholdPercentage: 50, resetTimeout: 30000, }); // Fails fast when circuit is open for (const user of users) { await breaker.fire(user).catch(handleFailure); }
Pitfall #9: Logging Sensitive Data
❌ Anti-Pattern
console.log('Request:', JSON.stringify(request)); // Logs API key, PII console.log('User:', user); // Logs email, phone
✅ Better Approach
const redacted = { ...request, apiKey: '[REDACTED]', user: { id: user.id }, // Only non-sensitive fields }; console.log('Request:', JSON.stringify(redacted));
Pitfall #10: No Graceful Degradation
❌ Anti-Pattern
// Entire feature broken if Perplexity is down const recommendations = await perplexityClient.getRecommendations(userId); return renderPage({ recommendations }); // Page crashes
✅ Better Approach
let recommendations; try { recommendations = await perplexityClient.getRecommendations(userId); } catch (error) { recommendations = await getFallbackRecommendations(userId); reportDegradedService('perplexity', error); } return renderPage({ recommendations, degraded: !recommendations });
Instructions
Step 1: Review for Anti-Patterns
Scan codebase for each pitfall pattern.
Step 2: Prioritize Fixes
Address security issues first, then performance.
Step 3: Implement Better Approach
Replace anti-patterns with recommended patterns.
Step 4: Add Prevention
Set up linting and CI checks to prevent recurrence.
Output
- Anti-patterns identified
- Fixes prioritized and implemented
- Prevention measures in place
- Code quality improved
Error Handling
| Issue | Cause | Solution |
|---|---|---|
| Too many findings | Legacy codebase | Prioritize security first |
| Pattern not detected | Complex code | Manual review |
| False positive | Similar code | Whitelist exceptions |
| Fix breaks tests | Behavior change | Update tests |
Examples
Quick Pitfall Scan
# Check for common pitfalls grep -r "sk_live_" --include="*.ts" src/ # Key leakage grep -r "console.log" --include="*.ts" src/ # Potential PII logging
Resources
Quick Reference Card
| Pitfall | Detection | Prevention |
|---|---|---|
| Sync in request | High latency | Use queues |
| Rate limit ignore | 429 errors | Implement backoff |
| Key leakage | Git history scan | Env vars, .gitignore |
| No idempotency | Duplicate records | Idempotency keys |
| Unverified webhooks | Security audit | Signature verification |
| Missing error handling | Crashes | Try-catch, types |
| Hardcoded config | Code review | Environment variables |
| No circuit breaker | Cascading failures | opossum, resilience4j |
| Logging PII | Log audit | Redaction middleware |
| No degradation | Total outages | Fallback systems |
Similar Claude Skills & Agent Workflows
skill-developer
Create and manage Claude Code skills following Anthropic best practices.
documentation-lookup
This skill should be used when the user asks about libraries, frameworks, API references, or needs code examples.
教程美化方案
使用 VitePress 和 Element Plus 组件美化教程,提升可读性和交互性
material-component-doc
用于 FlowGram 物料库组件文档撰写的专用技能,提供组件文档生成、Story 创建、翻译等功能的指导和自动化支持
langgraph-docs
Use this skill for requests related to LangGraph in order to fetch relevant documentation to provide accurate, up-to-date guidance.
openai-knowledge
Use when working with the OpenAI API (Responses API) or OpenAI platform features (tools, streaming, Realtime API, auth, models, rate limits, MCP) and you need authoritative, up-to-date documentation (schemas, examples, limits, edge cases).