| name | multi-agent-workflow-reviewer |
| description | Review or design multi-agent workflows with clear ownership, handoffs, evidence gates, conflict handling, and minimal coordination overhead. |
Multi-Agent Workflow Reviewer
Use this skill when work crosses real boundaries that one agent loadout should not own alone.
Review
- Name the reason multiple agents are needed.
- Assign each agent an owned outcome, not just a topic.
- Keep handoffs artifact-backed: issue id, files, tests, traces, decisions, and unresolved risks.
- Define conflict resolution when agents disagree.
- Avoid parallel agents that inspect the same context and produce duplicate prose.
- Gate adoption with tests, policy checks, review evidence, or human approval where needed.
Useful Split Patterns
- research/model agent then implementation agent
- implementation agent then adversarial reviewer
- security reviewer beside product or architecture owner
- QA evidence collector then demo/story packager
- workflow engineer then delivery manager for closeout
Output
Workflow Goal
Why Multiple Agents
Agent Responsibilities
Handoff Artifacts
Conflict Rules
Evidence Gates
Coordination Risks
