Perplexity Skill Pack

Claude Code skill pack for Perplexity Sonar API — AI-powered search with web-grounded answers and citations (30 skills)

What This Does

Gives Claude Code deep knowledge of the Perplexity Sonar API: the OpenAI-compatible endpoint at api.perplexity.ai that performs live web searches and returns cited, source-grounded answers. Skills cover the full lifecycle from first API call to production monitoring.

API Endpoint: https://api.perplexity.ai/chat/completions (OpenAI-compatible) Models: sonar | sonar-pro | sonar-reasoning-pro | sonar-deep-research Auth: Bearer token with pplx- prefixed API key SDK: Standard openai package with custom baseURL (no Perplexity-specific SDK)

Installation

/plugin install perplexity-pack@claude-code-plugins-plus

Skills Included

Standard Skills (S01-S12)

| Skill | What It Does | |-------|-------------| | perplexity-install-auth | Configure OpenAI client with Perplexity base URL and API key | | perplexity-hello-world | First search query with citation parsing and streaming | | perplexity-local-dev-loop | Dev workflow with fixture-based mocking and live test gating | | perplexity-sdk-patterns | Typed client wrapper, retry logic, citation formatter | | perplexity-core-workflow-a | Single-query search with citations and domain filtering | | perplexity-core-workflow-b | Multi-turn research sessions and batch query pipelines | | perplexity-common-errors | 401/402/429/400 error reference with real error JSON | | perplexity-debug-bundle | Diagnostic bundle: DNS, auth, model access, latency | | perplexity-rate-limits | Exponential backoff, PQueue rate limiting, token bucket | | perplexity-security-basics | API key rotation, PII query sanitization, domain restrictions | | perplexity-prod-checklist | Production readiness checklist with health check and fallback | | perplexity-upgrade-migration | Legacy model migration map (pplx-7b -> sonar) |

Pro Skills (P13-P18)

| Skill | What It Does | |-------|-------------| | perplexity-ci-integration | GitHub Actions with mocked unit tests and gated live tests | | perplexity-deploy-integration | Vercel edge function, Cloud Run with Redis cache, Docker | | perplexity-webhooks-events | Streaming SSE, batch queue pipeline, scheduled news monitor | | perplexity-performance-tuning | Model routing by complexity, query-aware cache TTL, streaming | | perplexity-cost-tuning | Model routing saving 60-70%, token limits, cache, budget tracking | | perplexity-reference-architecture | Search service + citation pipeline + research orchestrator |

Flagship Skills (F19-F24)

| Skill | What It Does | |-------|-------------| | perplexity-multi-env-setup | Dev/staging/prod configs with per-env model and key management | | perplexity-observability | Latency, citation quality, cost tracking, Prometheus alerts | | perplexity-incident-runbook | Triage script, decision tree, model fallback, postmortem template | | perplexity-data-handling | Query PII sanitization, citation validation, context management | | perplexity-enterprise-rbac | Per-team API keys, model policy gateway, domain restrictions | | perplexity-migration-deep-dive | Migrate from Google CSE/Bing to Perplexity with adapter pattern |

Flagship+ Skills (X25-X30)

| Skill | What It Does | |-------|-------------| | perplexity-advanced-troubleshooting | Layer diagnostics, citation stability, latency profiling | | perplexity-load-scale | k6 load tests calibrated for 50 RPM, capacity calculator | | perplexity-reliability-patterns | Circuit breaker, model fallback chain, stream timeout, stale cache | | perplexity-policy-guardrails | Content moderation, usage quotas, citation quality scoring | | perplexity-architecture-variants | Three blueprints: widget (<500/day), cached (5K), pipeline (5K+) | | perplexity-known-pitfalls | 10 real anti-patterns with BAD/GOOD code comparisons |

Quick Start

import OpenAI from "openai";

const perplexity = new OpenAI({
  apiKey: process.env.PERPLEXITY_API_KEY,
  baseURL: "https://api.perplexity.ai",
});

const response = await perplexity.chat.completions.create({
  model: "sonar",
  messages: [{ role: "user", content: "What are the latest AI developments?" }],
});

console.log(response.choices[0].message.content);
console.log("Sources:", (response as any).citations);

Key API Parameters

| Parameter | Values | Purpose | |-----------|--------|---------| | model | sonar, sonar-pro, sonar-reasoning-pro, sonar-deep-research | Model selection | | search_recency_filter | hour, day, week, month | Limit source freshness | | search_domain_filter | ["python.org"] or ["-reddit.com"] | Allow/deny domains (max 20) | | return_related_questions | true | Get follow-up suggestions | | return_images | true | Get relevant images (Tier-2+) | | stream | true | SSE streaming response |

License

MIT