II.
Workflow overview
Reference · liveworkflow:multi-agent-orchestration-review
Multi-Agent Orchestration Review overview
Reviews multi-agent system architectures before production deployment — auditing agent role definitions, inter-agent communication protocols, state graph correctness, tool-use permission boundaries, escalation and human-in-the-loop gates, cost-per-invocation budgets, and failure-mode handling across LangGraph/CrewAI/AutoGen-style orchestrations. Validates that agent loops terminate and outputs converge. Excludes individual agent prompt engineering.
Attributes
displayName
Multi-Agent Orchestration Review
workflowKind
governance
triggerType
event-driven
typicalCadence
per-deployment
complexity
cross-team
description
Reviews multi-agent system architectures before production deployment —
auditing agent role definitions, inter-agent communication protocols, state
graph correctness, tool-use permission boundaries, escalation and
human-in-the-loop gates, cost-per-invocation budgets, and failure-mode
handling across LangGraph/CrewAI/AutoGen-style orchestrations. Validates
that agent loops terminate and outputs converge. Excludes individual agent
prompt engineering.
Outgoing edges
applies_to_domain2
- domain:ml-ops·DomainMLOps
- domain:software-engineering·DomainSoftware Engineering
involves_role3
- role:ai-champion·RoleAI Champion
- role:staff-engineer·RoleStaff Engineer
- role:ml-engineer·RoleMachine Learning Engineer
performed_by_org_unit3
- org-unit:ai-enablement·OrgUnitAI Enablement
- org-unit:architecture-guild·OrgUnitArchitecture Guild
- org-unit:ml-team·OrgUnitML Team
requires_skill_area3
- skill-area:agentic-loops·SkillAreaAgentic Loops
- skill-area:tool-use·SkillAreaLLM Tool Use
- skill-area:context-management·SkillAreaLLM Context Management
triggers_responsibility2
- responsibility:review-architecture-changes·ResponsibilityReview architecture changes
- responsibility:ai-safety-guardrails·Responsibility
Incoming edges
follows_workflow1
- stack-profile:multi-agent-orchestration·StackProfile