Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →majiayu000/openai-agent-sdk-integration
This skill implements AI agents with MCP (Model Context Protocol) server integration using the OpenAI Agents SDK. It covers agent creation, tool definitions, MCP server connections (stdio, HTTP, SSE), tool filtering, caching, handoffs, guardrails, and multi-agent orchestration patterns.
Passed 23 of 23 test cases with the skill, and 5 of 23 without it.
and 22 more in the full report
18 fixed5 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 18 test cases the skill fixed. Same model, same prompt, with and without the skill.
I am writing a non-async Python script that executes an agent request synchronously. Which entry point method on Runner should be called?
| Checked | Without skill | With skill |
|---|---|---|
| Calls Runner.run_sync() | Missed | Passed |
The instructions your model reads when the skill loads.
This skill provides comprehensive guidance for building AI agents using the OpenAI Agents SDK with MCP (Model Context Protocol) server integration.
| Source | Gather |
|---|---|
| Codebase | Existing agent patterns, MCP server implementations, tool definitions |
| Conversation | User's specific requirements, transport type (HTTP/stdio), tool needs |
| Skill References | Domain patterns from references/ (SDK patterns, MCP primitives, examples) |
| User Guidelines | Project-specific conventions, authentication requirements |
| Primitive | Purpose | Control Model |
|---|---|---|
| Agents | LLMs with instructions and tools | Framework-controlled |
| Handoffs | Delegation between agents | Agent-controlled |
| Guardrails | Input/output validation | Developer-controlled |
| Primitive | Purpose | Control Model |
|---|---|---|
| Tools | Functions AI can invoke | Model-controlled |
| Resources | Read-only data access | Application-controlled |
| Prompts | Pre-crafted instructions | User-controlled |
from mcp.server.fastmcp import FastMCP
mcp = FastMCP(
name="my-mcp-server",
test cases passed, vs 5 / 23 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.