Skip to main content
Source

This page is generated from devops-agent/eks-security/references/porting-notes.md. Edit the source, not this page.

Porting Notes — eks-security

This file documents the differences between the Claude Code version and the DevOps Agent port. It is for maintainers, not for the agent to read during execution.

Staleness check: the table below describes the upstream skill at a point in time and can drift as skills/eks-security/ evolves. Re-verify each row against upstream when materially changing either copy, and update the date here. Last verified: 2026-08-23 (round-4/5 currency + precision hardening re-verified across both copies — AM→Config conformance packs, exact pack names, PCI Req 3.5.1.1, audit-logging retention, evals; the 3 per-regime deep files were ported 2026-08-19; other rows carried forward unchanged from 2026-07-17).

Differences from Claude Code Version

AspectClaude Code versionDevOps Agent version
Execution modelInteractive — asks 8 discovery questions conversationallyAutonomous with HARD STOP gates — proceeds if context is sufficient, stops only for critical missing items
Discovery8 interactive questions before any recommendation3 mandatory context gates (compliance regime, workload sensitivity, OS/AMI preference); 5 additional context items gathered opportunistically
Tool accessUses Bash, kubectl, AWS CLI via MCP server for live cluster inspectionUses AWS APIs and Kubernetes APIs available in the Agent Space (read-only)
EscalationReferences internal SpecReq / Specialist processesRecommends engaging AWS Professional Services or Solutions Architects
Skill routingRoutes to sibling skills (eks-genai, eks-build, eks-design)Self-contained; notes alternative guidance domains without routing
Script executionCan run kube-bench, generate shell commandsAdvisory only — recommends commands for the user to execute
MCP dependenciesReferences eks-mcp-server for live dataNo MCP dependencies; uses Agent Space APIs directly
Auto Mode security referencereferences/auto-mode-security.md — security facts (node OS, IMDSv2, shared-responsibility split)Identical / in sync; security facts are launch-agnostic, only execution-model framing differs — edit both copies together.
Compliance references3 per-regime deep files (compliance-hipaa/pci/soc2.md), each with a control-mapping table + 30/60/90 + shared-responsibility table, plus router bullets + worked scenarios in compliance-regimes.mdAt full parity — the same 3 per-regime deep files (compliance-hipaa/pci/soc2.md) are now ported (byte-identical; their only outbound links are to compliance-accelerators.md + compliance-regimes.md + engagement-and-response.md, all present in the port), each carrying the control-mapping table + 30/60/90 + shared-responsibility table. They are wired identically to the skill: the SKILL.md Compliance-Regime Scope paragraph + the "How to Use the References" table link to all three, and compliance-regimes.md carries the router intro + per-regime deep-dive links. Keep the compliance facts in sync across both copies.