Ac Multi Agent Orchestration

The craft of running more than one agent on one job: deciding when to split at all, designing roles and scopes, budgeting context between agents, writing delegation prompts, verifying agent products independently, recovering failed agents, and keeping a ledger of what the loop costs. Seven leaves: when-to-split, role-and-scope, context-budget, delegation-prompts, verification-loops, failure-and-recovery, ledger-and-leverage. Use when asking 'orchestrate multiple agents', 'multi-agent workflow', 'split this work across agents', 'design an agent team', 'delegate to subagents', 'should I use more than one agent'. Not for handoff artifacts between people and sessions (ac-handoff), triaging one human's inbound queue (ac-inbound-triage), or prompting a single model (ac-prompt-engineer).

jiayaoqijia Updated 81 repo stars

File contents

jiayaoqijia/cryptoskill commit 10cf155307

Frequently asked questions

npx get-skill add jiayaoqijia/ac-multi-agent-orchestration