LPMS Change Impact Guard

LPMS Change Impact Guard helps developers understand the real impact of a change before they start editing a complex Learning Plan Management System. The problem is that a seemingly small request can cross many parts of LPMS: approval pages, reviewer roles, comments, reference records, workflow stages, server persistence, and regression tests. Developers can easily miss an affected rule, change a workflow by accident, or update only the interface while leaving the underlying behavior inconsistent. For this MVP, a developer enters one realistic change request: require the Director of Libraries to leave an actionable comment on a specific reference before returning a learning plan. The Guard produces an evidence-based report that identifies affected routes, source files, roles, workflow transitions, existing tests, a safe implementation order, assumptions, and explicit “do not touch” boundaries. It is deliberately deterministic and local. It does not invent files, modify source code, change approval data, create database records, or call a runtime AI service. Unsupported requests return “Manual review required” instead of guessing. The tool turns an existing LPMS codebase into a safer AI-assisted development workflow: people can inspect the likely blast radius, approve or revise a plan locally, and export Markdown or JSON evidence before implementation begins.