execute

What problem does it solve? Coordinating multi-step coding tasks across coder, reviewer, and test agents often leads to skipped verification steps, unverified fixes, and premature commits. This protocol enforces a strict execution pipeline where every task passes through automated gates and agent reviews before completion. ## Core Features & Use Cases - Structured Gate Pipeline: Runs diff, syntax check, placeholder scan, imports audit, lint, build check, and a parallel pre_check_batch (lint, secretscan, SAST, quality budget) for every task. - Coder Retry Protocol: Returns structured gate-failure rejections to the coder agent with exact error output, resuming at the failed step instead of restarting. - Regression and Drift Detection: Runs graph-scoped regression sweeps and test-drift checks on drift-prone changes like CLI behavior, parsing, and public contracts. - Completion Evidence Checklist: Requires a filled task-completion gate checklist from actual tool output before any task can be marked complete or committed. - Use Case: An architect agent executing a phased implementation plan uses this protocol to delegate each task to a coder, enforce all QA gates, and only commit after reviewer approval and passing tests. ## Quick Start Load the execute protocol and run the current task through the full gate pipeline, delegating fixes to the coder agent on any gate failure.