team-planex

Task decomposition and sequential execution are orchestrated by a plan-and-execute pipeline that generates actionable implementation tasks from requirements and verifies them sequentially. It delegates planning to a planner worker and implementation to an executor while a coordinator manages progression.

An issue-based plan-and-execute development pipeline is a workflow that accepts issue IDs, text requirements, or plan files to create PLAN-* tasks, then generates corresponding EXEC-* tasks per issue. The coordinator spawns team-worker agents to implement each solution with verification and cross-task consistency tracking.

Session state and solution artifacts are persisted by creating a reproducible session directory that stores task analysis, solution artifacts, and wisdom. This enables cross-task consistency while the coordinator handles callbacks for task readiness, progress, and blocker reporting throughout the execution lifecycle.

Codex and gemini execution methods are supported with interactive or auto selection, including fallback behavior and retry-on-failure rules. The pipeline allows you to choose the executor dynamically at runtime while handling task readiness and completion callbacks during the spawn-and-stop progression.

A mandatory SKILL.md router definition is required to coordinate planner and executor workers. The pipeline requires this router setup to spawn team-worker agents, generate EXEC-* tasks at runtime from PLAN-* inputs, and maintain session state tracking with progress and blocker reporting.

Running team-planex on issue IDs requires asking the AI to process identifiers like ISS-20250101-123456. The coordinator decomposes these requirements into PLAN-* tasks, generates corresponding EXEC-* tasks, and executes them sequentially with verification while persisting session artifacts for reproducibility.