A reference for anyone building multi-step agents in Claude Code and trying to decide where the workflow logic should sit.

You are building an agent that does multiple things. Maybe it is a data pipeline (extract, transform, validate, load). Maybe it is a code review agent (lint, test, security scan, generate report). Maybe it is something custom for your team with four or five distinct steps.

February 2026

This guide describes how to enable EKS Pod Identity for projects using a GitOps approach. The pipeline processes YAML request files and configures IAM trust policies and Pod Identity associations.

When a project is provisioned, the namespace and service account are created immediately. However, due to IAM permission restrictions, the trust policy update and Pod Identity association must be handled by an infrastructure pipeline with elevated permissions.

This questionnaire is intended to understand your current technology, data, and analytics ecosystem in order to assess readiness for advanced Future of Sales capabilities such as:

• account-level intelligence (Account 360)

Understanding the Protocol vs Platform Distinction

MCP (Model Context Protocol) and AWS AgentCore are complementary technologies, not alternatives:

• MCP = An open communication protocol (like HTTP for AI agents)

Discovery Strategy: Analyzing Legacy Applications

What We Need to Discover

Based on info:

1. Java monolith with stored procedures (8 applications)
2. AppShell integration - micro-frontend architecture exists
3. React target - frontend framework confirmed
4. Backend API strategy - REST? GraphQL? Hybrid? (Your question pending)
5. Domain boundaries - where to draw subgraph lines

Entry point where users select their persona/role to access the appropriate studio environment.

┌─────────────────────────────────────────────────────────────────────────────────────┐
│  AI AGENTIC PLATFORM                                     [User: John]  [Settings] ⚙ │

Leadership Overview | Comprehensive Framework | 7-Layer Architecture

Enterprises pursuing agentic AI face critical gaps that prevent successful adoption:

Banking & Financial Services - Production Ready

1. Architecture Overview
2. Layer 1: Experience Layer

                                    ┌────────────────────────────────────────────┐
                                 │               Business Personas            │
                                 │   • Ops Analysts • Business Units         │
                                 │   • Helpdesk • Risk & Compliance          │
                                 └───────────────────────▲────────────────────┘
                                                         │
                                        Thin Control Plane / Portal
                                                         │