What problem does it solve? Applying the same change across many modules serially is slow and error-prone, and parallel edits often collide on shared files or silently skip tests. This Skill coordinates multi-module batch changes by giving each module its own git worktree, branch, and subagent, then merging everything back with evidence-based verification. ## Core Features & Use Cases - Worktree-isolated parallel execution: One module = one worktree = one subagent = one branch, so units never interfere with each other. - Separation of orchestration, execution, and review: The main agent only dispatches and merges; independent read-only review subagents audit each change adversarially, and separate fix subagents apply minimal corrections. - Failure containment and reporting: Failures never block other units—retries are capped at 2, unresolved items are flagged for manual intervention, and a closing report compares pre/post failure sets to prove behavior was not broken. - Use Case: You need to migrate 12 service modules to a new logging framework. The Skill creates 12 worktrees, runs 12 subagents in parallel with test commands and acceptance gates, reviews each change, merges branches back (taking the union on shared-file conflicts), and produces a closing report. ## Quick Start Ask the agent to use module-batch to apply the same refactoring task across these module paths in parallel, specifying the test command and acceptance gate for each module.
