Lead agent in the review swarm In this channel's material, a lead agent is the coordinating agent in a multi-agent workflow: it collects the findings produced by other agents and turns them into a consolidated report of what needs to be done [^7d2a9b32-b281-5828-9eb1-bf812129ae58@426]. Where it appears The clearest example is Ryan Nystrom's swarm review skill, a multi-agent verification pass that runs before he ever reviews code himself. The workflow divides the change set into domains (for example frontend and backend), delegates each domain to reviewer roles — a correctness reviewer hunting for errors and a code maintenance reviewer looking for reuse, existing patterns, and things that break at scale — and runs every pass on two models (GPT and Opus). At the end of the pipeline, the lead agent collects all findings from every reviewer pass and writes them up as a report of required fixes [^7d2a9b32-b281-5828-9eb1-bf812129ae58@426]. How to use the loop The swarm iterates until it's good: the agent repeats the review cycle until all findings are resolved [^7d2a9b32-b281-5828-9eb1-bf812129ae58@426]. - A lead agent is the coordinator: it gathers outputs from specialized reviewer agents and produces the final report of what needs to be done.- In the swarm review skill, the lead agent sits at the end of a pipeline of domain-split, role-based, multi-model reviewer passes.- Its report feeds an iterate until it's good loop that repeats until the findings are resolved.