Skip to main content

Workstreams

WorkstreamWhat it isLeaders and Core Contributors
Agent Lifecycle ManagementManaging agent lifecycle on the platform via the operator — deploy, hibernate, wake, scaleJarek Cwiklik, Paolo Dettori
Agent RuntimeAgent runtime, proxyHai Huang, Ed Snible
Audit TrailSecurity audit trail — full record of autonomous agent actions for complianceParul Singh
Authorization and Policy ManagementScoped permissions with runtime policy enforcement that limit what an agent can doRong Chang, Maia Iyer, Alan Cha, Hai Huang
BenchmarkingAgent benchmarking and test infrastructureKelly Abuelsaad
GuardrailsContent safety and compliance policies for agents making autonomous decisionsEvaline Ju
Identity Management (Keycloak)Identity and access managementAlan Cha, Maia Iyer
Intent Based Access ControlIntent based access controlHai Huang
MCP GatewayUnified front door for MCP servers — tool routing, auth, token exchangeHai Huang, Ed Snible
Network SecuritySecure network architecture and isolation between agents and external systemsHai Huang
Observability & Token Cost ManagementTracing, token cost attribution, and auto-instrumentationEvaline Ju, Ilya Kolchinsky
Resiliency & ConsistencyEnsuring environment consistency and preventing corruption from agent actions, inconsistent statesSahil Suneja
SandboxingIsolated execution environments for code dev, AI-generated code, and MCP toolsLadislav Smola
State ManagementState, context, and session management — persistent state so agents retain memory across restarts and sessionsMaia Iyer, Evaline Ju, Alan Cha, Kelly Abuelsaad
Workload Identity and SecurityZero-trust auth and identity management for agents, tools, and users — cryptographic workload identity via SPIFFE/SPIREMariusz Sabath, Maia Iyer, Alan Cha, Morgan Foster, Akram Ben Aissi