ag-validar-execucao

You can validate a SPEC or task_plan.md by spawning a read-only subagent that compares every planned item against the codebase. It reports totals of done, partial, and missing items to verify implementation completeness without modifying the source.

Automated spec validation is the process of spawning a read-only agent to independently verify that every requirement listed in a SPEC or task_plan.md is actually implemented in the codebase, reporting done, partial, and missing totals without modifying source files.

Yes, you can spawn a read-only validation agent to compare a task_plan.md against the codebase. It identifies done, partial, and missing items to ensure complete implementation without modifying the source code.

Yes, execution plan validation works by applying a read-only agent to either a SPEC.md or task_plan.md file. It identifies and validates every planned item against the codebase to ensure complete implementation, reporting totals and gaps without modifying source.

The main limitation of a read-only validation agent is that it cannot fix missing or partial implementations. It only compares plan items with the code to report completeness totals and gaps, requiring you to manually address any identified issues afterward.

The best way to verify all planned items in a SPEC are implemented is to spawn a read-only validation agent. It independently compares every item against the codebase and reports done, partial, and missing totals to ensure complete implementation without altering the code.