maestro-plan

What problem does it solve?

Maestro-plan automates the labor-intensive process of exploring a codebase from multiple angles and turning those findings into a verified, actionable execution plan so teams don't manually synthesize discoveries, dependencies, and task breakdowns.

Core Features & Use Cases

• CSV wave pipeline: Two-phase workflow where Wave 1 runs parallel exploration agents and Wave 2 runs a sequential planning agent that emits plan.json and TASK-*.json.
• Parallel exploration + sequential planning: Uses spawn_agents_on_csv to scale exploration with concurrency controls, then consolidates findings for deterministic plan generation and validation.
• Plan validation and guardrails: Produces grep-verifiable convergence criteria, read_first requirements, discovery append-only board, gap-mode support, and a revision loop to ensure plan quality.
• Use Case: Quickly generate a phased execution plan for a development phase by exploring architecture, implementation patterns, integration points, and risks, then producing task definitions for execution.

Quick Start

Invoke maestro-plan for the target phase with appropriate flags (for example: run maestro-plan for phase 3 with --auto and --concurrency 4 to produce plan.json and TASK files).