Speak Skill Pack

Claude Code skill pack for Speak AI language learning platform integration (24 skills)

Installation

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

About Speak

Speak is an AI-powered language learning platform that gets users speaking from day one. Built on OpenAI's GPT-4o and Realtime API, Speak provides real-time pronunciation feedback, adaptive AI tutoring, and conversation practice without a live tutor. Backed by OpenAI Startup Fund, Founders Fund, and Y Combinator. Valued at $1B+ (Series C, 2024).

Key facts:

• 14+ supported languages (Korean, Spanish, Japanese, French, German, and more)
• GPT-4o for conversation generation and personalized feedback
• OpenAI Whisper + proprietary models for speech recognition
• Realtime API for sub-second response times in live roleplays
• Proficiency graph system tracks learner knowledge state

Skills Included

Standard Skills (S01-S12)

| Skill | Description | |-------|-------------| | speak-install-auth | Install SDK, configure API keys, set up speech pipeline | | speak-hello-world | First AI tutoring session with pronunciation scoring | | speak-local-dev-loop | Mock tutors, audio fixtures, debug mode | | speak-sdk-patterns | Session manager, audio preprocessor, retry, progress tracker | | speak-core-workflow-a | AI conversation practice with adaptive feedback | | speak-core-workflow-b | Pronunciation drills with phoneme-level analysis | | speak-common-errors | Error codes, diagnostic commands, recovery patterns | | speak-debug-bundle | Auth check, audio validator, network diagnostics | | speak-rate-limits | Throttled client, batch queue, tier limits | | speak-security-basics | API key security, audio privacy, COPPA/FERPA | | speak-prod-checklist | Audio pipeline, monitoring, compliance checklist | | speak-upgrade-migration | SDK upgrades, API version migration |

Pro Skills (P13-P18)

| Skill | Description | |-------|-------------| | speak-ci-integration | GitHub Actions with mocked API tests | | speak-deploy-integration | Deploy to Vercel, Cloud Run, containers | | speak-webhooks-events | Lesson completion and progress events | | speak-performance-tuning | Audio preprocessing, caching, connection pooling | | speak-cost-tuning | Usage monitoring, tier optimization | | speak-reference-architecture | Client, API gateway, assessment engine, progress store |

Flagship Skills (F19-F24)

| Skill | Description | |-------|-------------| | speak-multi-env-setup | Dev/staging/prod with separate keys and mock modes | | speak-observability | API health, assessment latency, score distributions | | speak-incident-runbook | Outage triage, offline fallback, recovery | | speak-data-handling | Audio privacy, GDPR/COPPA compliance | | speak-enterprise-rbac | SSO, teacher/student roles, class management | | speak-migration-deep-dive | Platform migration, progress import |

Quick Start

import { SpeakClient } from '@speak/language-sdk';

const client = new SpeakClient({
  apiKey: process.env.SPEAK_API_KEY!,
  appId: process.env.SPEAK_APP_ID!,
  language: 'es',
});

// Start conversation practice
const session = await client.startConversation({
  scenario: 'ordering-food',
  language: 'es',
  level: 'intermediate',
});
console.log('Tutor:', session.firstPrompt.text);

// Assess pronunciation
const result = await client.assessPronunciation({
  audioPath: './recording.wav',
  targetText: 'Hola, como estas?',
  language: 'es',
  detailLevel: 'phoneme',
});
console.log(`Score: ${result.score}/100`);

Resources

• Speak Website
• Speak GPT-4 Blog
• OpenAI Realtime API
• Live Roleplays Feature

License

MIT