Faster Together: How I Use AI and MCP to Build Better Connectors
summary
Builds a working Apollo connector for a public random-user REST API using Claude plus the Apollo MCP tooling, auto-generates schema and mapping, composes and runs it with rover supergraph run, and creates YAML unit tests that pass when executed with rover connector test. Shows new mapping features in connectors 0.3 — the ? and ?? operators and arrow methods — and demos an orchestration flow that partitions many OpenAPI specs into data domains, runs four parallel agents, composes the supergraph, and fixes federation errors automatically with Apollo MCP Server.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:05AI coding tools and LLM assistance
- 2:31Apollo connectors and MCP server
- 3:36Demo: Building connector with AI
- 10:59New language features in Connect 0.3
- 12:22Demo: Enhancing connectors with AI
- 18:43AI-driven connector testing
- 19:03Future of AI-orchestrated APIs
- 24:25Future of AI and MCP tools
- 25:27Q&A
- 27:13Q&A: Cross-graph entity references
- 30:21Q&A: Performance guardrails for LLM connectors
- 32:29Q&A: Connectors vs traditional subgraphs
- 36:02Q&A: Conditional routing in connectors
- 38:28Q&A: Decomposing large schemas with MCP
- 40:37Q&A: Connectors vs REST APIs and hosting
related talks
26mOrchestrating APIs for LLMs - Agentic Workflows for GraphQL
01h 23mWorkshop: Apollo Router & MCP: A Modern Agentic Development Approach
29mBuilding MCP Apps With GraphQL Patterns You Already Know
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
30mSchema Design Patterns: Leveraging Existing REST APIs for Rapid GraphQL
28mTalk to your data without coding agentic AI and Claude with the MCP protocol
26mScaling Your AI Agents with Open Source: API ➔ No-Code MCP & Observability wit…
26mAsk, Don't Fetch: Why GraphQL + MCP Is the Native Language of AI Agents