sdd:implement

Automated task implementation uses LLM-as-Judge verification to ensure high-quality code by coordinating sub-agents in iterative fix-verify cycles. It removes manual orchestration by applying configurable quality thresholds to multi-step development workflows.

LLM-as-Judge verification is an automated quality assurance mechanism that evaluates code artifacts against configurable thresholds. It uses multiple judges in iterative fix-verify cycles to orchestrate incremental refinements and ensure high-quality outputs.

You can resume incomplete task workflows by leveraging the skill's flexible workflow modes, which support continuing from incomplete steps and applying incremental refinements based on detected git changes. Optional human-in-the-loop checkpoints can be configured for manual validation.

Automated task orchestration supports optional human-in-the-loop checkpoints alongside continue and refine modes. This allows developers to manually validate code artifacts during multi-step workflows before meeting the configurable quality thresholds.

The best way to orchestrate complex code generation is using automated sub-agents with LLM-as-Judge verification and configurable quality thresholds. This approach handles incremental refinement through iterative fix-verify cycles and git-based change detection.

You should not use automated LLM-as-Judge task implementation for simple, single-step modifications that lack multi-step orchestration requirements or configurable quality thresholds. It is designed for complex workflows requiring incremental refinement and git-based change detection.