build-step

What problem does it solve?

Coordinate the implementation, test-driven validation, and peer review of individual epic steps so agentic teams can make reliable, auditable progress without human micromanagement.

Core Features & Use Cases

• Per-step orchestration: Parse an implementation plan, initialize step state, and iterate through pending tasks for a single epic.
• Agentized dev + review loop: Spawn fresh developer and reviewer sub-agents via the Task tool for each step to maintain clean contexts and enforce review gates.
• Robust safety and recovery: Enforce test immutability via TDD baselines, track circuit-breaker signals, and optionally escalate to automated replanning to correct or justify failing tests.
• Use Case: Use from an execute-plan orchestrator to implement an epic by iteratively running development attempts, automated tests, reviews, and controlled replanning until completion.

Quick Start

Run the build-step skill from the execute-plan orchestrator with epic_name, build_dir, plan_path, epic_spec_path, tdd_baseline_tag, and mode to process the next epic step-by-step.