openai-agents

✓Verified·Scanned 2/18/2026

This skill provides SDK documentation, templates, and code samples for building OpenAI Agents. It sets/uses OPENAI_API_KEY, includes external requests (e.g., https://api.example.com/search?q=${query}), and contains shell/CLI commands such as npm view @openai/agents version.

by jezweb·v10a1f16·148.5 KB·361 installs
Scanned from main at 10a1f16 · Transparency log ↗
$ vett add jezweb/claude-skills/openai-agents

OpenAI Agents SDK Skill

Complete production-ready skill for building AI applications with OpenAI Agents SDK (JavaScript/TypeScript).

Auto-Trigger Keywords

This skill auto-triggers when you mention:

Core Concepts:

  • openai agents
  • openai agents sdk
  • openai agents js
  • openai agents typescript
  • @openai/agents
  • @openai/agents-realtime
  • agent sdk
  • openai agentic
  • agentic workflows
  • ai agents framework

Features:

  • multi-agent
  • agent handoffs
  • agent delegation
  • agent orchestration
  • agent routing
  • tool calling agents
  • function calling agents
  • agent tools
  • agent guardrails
  • input guardrails
  • output guardrails
  • structured output agents
  • zod schema agents
  • streaming agents
  • realtime agents
  • voice agents
  • openai realtime
  • openai voice
  • human-in-the-loop agents
  • hitl patterns
  • agent approval

Frameworks:

  • cloudflare workers agents
  • nextjs agents
  • hono agents
  • react voice agents
  • voice ui react

Patterns:

  • llm orchestration
  • code-based orchestration
  • agents as tools
  • parallel agents
  • agent evaluation
  • agent feedback loops
  • triage agent
  • specialist agents

Errors:

  • zod schema type error agents
  • mcp tracing error
  • maxturnsexceedederror
  • toolcallerror agents
  • guardrail error
  • schema mismatch agents

Use Cases:

  • customer service agents
  • support ticket agents
  • billing automation agents
  • technical support automation
  • documentation assistant
  • voice assistant
  • conversational ai
  • ai workflow automation
  • automated research agents
  • content generation agents

What This Skill Provides

Text Agents

  • ✅ Basic agent creation with tools
  • ✅ Multi-agent handoffs (triage pattern)
  • ✅ Structured outputs with Zod
  • ✅ Streaming responses
  • ✅ Input/output guardrails
  • ✅ Human-in-the-loop patterns
  • ✅ Parallel execution

Realtime Voice Agents

  • ✅ Voice agent setup (WebRTC/WebSocket)
  • ✅ Browser session management (React)
  • ✅ Voice agent handoffs
  • ✅ Audio I/O handling
  • ✅ Transport options

Framework Integration

  • ✅ Cloudflare Workers (experimental)
  • ✅ Next.js App Router
  • ✅ Hono framework
  • ✅ React components

Error Prevention

  • ✅ 9+ documented errors with workarounds
  • ✅ Retry patterns
  • ✅ Fallback strategies
  • ✅ Debugging techniques

Templates Included

17 Production Templates:

  1. Basic agent with tools
  2. Multi-agent handoffs (triage)
  3. Structured output (Zod)
  4. Streaming events
  5. Input guardrails
  6. Output guardrails
  7. Human approval (HITL)
  8. Parallel execution
  9. Realtime voice agent
  10. Browser voice client (React)
  11. Voice agent handoffs
  12. Cloudflare Workers
  13. Hono integration
  14. Next.js API route
  15. Next.js realtime endpoint
  16. Error handling
  17. Tracing/debugging

References Included

5 Comprehensive Guides:

  1. agent-patterns.md - LLM vs code orchestration, agents as tools, parallel patterns
  2. common-errors.md - 9 errors with GitHub issue links and solutions
  3. realtime-transports.md - WebRTC vs WebSocket comparison
  4. cloudflare-integration.md - Experimental Workers support, limitations
  5. official-links.md - Documentation, GitHub, npm, examples

Metrics

  • Token Savings: ~60% (12k → 5k tokens)
  • Errors Prevented: 9 documented issues
  • Templates: 17 production-ready
  • Package Version: @openai/agents@0.2.1
  • Last Verified: 2025-10-26

Quick Start

# Install
npm install @openai/agents zod@3

# Set API key
export OPENAI_API_KEY="your-key"

# Create agent
import { Agent, run } from '@openai/agents';

const agent = new Agent({
  name: 'Assistant',
  instructions: 'You are helpful.',
});

const result = await run(agent, 'Hello!');
console.log(result.finalOutput);

When to Use

✅ Use when:

  • Building multi-agent workflows
  • Creating voice AI applications
  • Need tool calling with validation
  • Require guardrails for safety
  • Implementing human approval gates
  • Deploying to Cloudflare/Next.js

❌ Don't use when:

  • Simple OpenAI API calls
  • Non-OpenAI models only
  • Massive scale voice (use LiveKit)

Production Tested

This skill has been tested in production environments and includes:

  • Error handling for all major failure modes
  • Security best practices (API key protection)
  • Cost optimization techniques
  • Debugging and tracing patterns

Official Resources

Skill Version: 1.0.0 SDK Version: 0.2.1 License: MIT Author: Jeremy Dawes (Jezweb)