perfectcode-zen-plan

Multi-agent feature planning uses parallel blueprints from code, performance, devops, and test engineers, followed by senior architect arbitration to resolve conflicts and produce a final structured plan stored in .opencode/plans.

Multi-agent planning assigns parallel blueprint generation to specialized engineering agents, then escalates high-stakes conflicts or design decisions to senior architect and security agents for arbitration before producing the final plan.

A complete feature plan should include sections for summary, scope, architecture, implementation tasks, testing strategy, migration path, rollout plan, and open questions to guide downstream development work.

To start the planning workflow, provide a kebab-case feature name and initial requirements, then trigger the workflow to automatically generate a complete plan saved to .opencode/plans/<feature-name>.md.

Senior architect arbitration should be used during feature planning when conflicts arise between parallel engineering blueprints or when high-stakes architectural, database, or security decisions need resolution.

Multi-agent feature planning operates without external dependencies, coordinating internal agents for blueprint production and arbitration to generate comprehensive plans independently.