SUMMARY.md

Summary

Introduction

Getting started

• Quickstart
• Getting started
• CLI reference
• Use cases
  • Exec-brief walkthrough

Architecture

• Architecture overview
• Architecture diagrams
• Runtimes
  • Runtime contract (BYO)
• A2A gateway (architecture)
• AGT boundary
• Per-sandbox identity (Entra Agent ID) — deep dive
  • Runtime token flow
  • Security alignment
  • Mesh trust design
• Multi-tenant model
• Egress proxy

Security

• Security overview
• Feature maturity & status
• Control mapping
• STRIDE × trust boundaries
• Red team playbook
• CRD trust model
• Security validation
• Supply-chain posture & OpenSSF Scorecard
• MCP top-10
• Upstream alignment

Agent capabilities

• kars OpenClaw plugin
• kars Hermes plugin
• @kars/mesh plugin (local OpenClaw)
• Channels & external plugins
• MCP servers
• Operator TUI
• Permissions model
• Per-sandbox identity (Entra Agent ID)
• Examples catalogue

Runbooks

• Autonomous SRE operator
• Hermes runtime — troubleshooting

Operations

• Operations overview
• A2A gateway (operations)
• BYO strict mode
• Branch protection
• Chaos tier
• GitOps
• Helm packaging
• Image versioning
• Upgrades & rollback
• Secret rotation
• Supply chain

API & policy

• CRD reference
• KarsEval (operator guide)
• Lifecycle & reconciliation
• Conditions
• Policy canonical format

Blueprints

• Index
• Developer inner loop
• Local k8s dev loop
• Enterprise self-hosted
• Managed public offload
• Cross-org federation
• Sovereign / air-gapped

Roadmap & ADRs

• Roadmap
• ADR index
  • ADR-0001: A2A ingress front edge
  • ADR-0002: Inference endpoint sourcing