What problem does it solve? Plans get written and then quietly ignored during implementation, and green tests do not prove the plan's contracts survived. This Skill runs an approved plan end to end, then verifies the code against the plan's spec, repo standards, and named contracts so drift gets caught instead of shipped. ## Core Features & Use Cases - Wave-based execution: Groups plan tasks into parallel or sequential waves, assigns each a capability tier (routine, standard, deep), and delegates to subagents with scoped briefing packs. - Plan verification: Checks spec compliance, repo standards from AGENTS.md and CLAUDE.md, and every contract and failure mode the plan named, reading the actual code rather than trusting subagent reports. - Structured reporting: Produces verification findings with a resolution gate, marks the plan file as EXECUTED, and offers follow-ups like code-review, cleanup, or publishing a draft branch. - Use Case: After approving an implementation plan, tell the agent to execute it; it dispatches the tasks, verifies the result against the plan, and reports any gaps before you commit. ## Quick Start Approve the plan I just showed you and execute it, verifying the implementation against the plan when finished.
