Skip to content

MCP Gateway Masterclass

Build governed, production-grade agentic AI

Hands-on labs culminating in a capstone where a CrewAI agent securely calls a Langflow tool via the MCP Gateway.

Home

Python 3.11+ FastAPI MkDocs Material


Welcome

The MCP Gateway Masterclass is a hands-on workshop for building enterprise-grade agentic AI. You’ll bring up an MCP Gateway, register tools, add guardrails, and finish with a CrewAI + Langflow capstone running end-to-end.

Format at a glance

2 days × (4h theory AM + 4h labs PM). All examples use port 4444 for the gateway to build muscle memory.


Quick Start

python3 -m venv .venv && source .venv/bin/activate
pip install -U mcp-contextforge-gateway
mcpgateway --host 0.0.0.0 --port 4444
curl -s http://localhost:4444/health | jq .
export MCPGATEWAY_BEARER_TOKEN=$(python3 -m mcpgateway.utils.create_jwt_token \
  --username [email protected] --exp 10080 --secret my-test-key)
mcp --server http://localhost:4444 tools list

Workshop Map

  • Syllabus

    ^^^

    Two days from quickstart to production patterns, with checklists for each block. Open »

  • Part I — Foundations

    ^^^

    MCP, gateway architecture, serving patterns, security & observability. Open »

  • Part II — Day-1 Labs (with solutions)

    ^^^

    Register servers, use clients, build wrappers, and enable rate-limits. Open »

  • Part III — Capstone Theory

    ^^^

    Design: Langflow ↔ Adapter ↔ Gateway ↔ CrewAI. Policies, RBAC, tracing. Open »

  • Part IV — Capstone Build (with solutions)

    ^^^

    Implement the adapter, run the CrewAI agent through the gateway, add guardrails & RBAC, capture traces. Open »

  • Appendices

    ^^^

    Docling + watsonx.ai RAG via the gateway, verified commands, API cheatsheets, troubleshooting, instructor RoS. Open »


Architecture at a Glance

flowchart LR
  A[CrewAI Agent] --> B(MCP Gateway)
  B --> C[Adapter/Server]
  C --> D[Langflow API\n/run flow]
  B -->|Guardrails: rate-limit, PII/secrets, RBAC| B
  B --> E[(Logs & Traces)]