task-validation-loop

You can validate your task list by running an autonomous verification loop that checks alignment with PRD, architecture, and design system. It launches parallel validators to assess task completeness and aggregates results to guide iterative fixes.

Multi-agent validation is an autonomous process that launches three parallel validators to assess task completeness against project specs. It aggregates verdicts and iteratively re-runs changed validators to ensure full alignment before implementation.

To automate task alignment in a monorepo, use a validation loop that resolves the correct specs path automatically. It evaluates your tasks.md and product_specs directories against architecture specs to produce a combined verdict.

Yes, task validation requires a complete task list, PRD, and architecture as prerequisites. The verification loop needs these defined specs in directories like tasks/ and product_specs/ to compare task completeness against project requirements.

The best way to fix incomplete task lists is launching parallel validators that assess alignment with PRD, architecture, and design system. The loop aggregates results, iterates fixes, and re-runs only changed validators to satisfy all criteria.

Task validation re-runs only changed validators to efficiently satisfy all criteria after iterative fixes. This optimization prevents redundant checks on unchanged specs while focusing on resolving the specific validation failures identified in the combined verdict.