What problem does it solve? Monorepo teams struggle to see whether their actual package dependency graph matches the architecture they intended. This Skill extracts the real runtime dependency graph from workspace manifests and renders it as a self-contained HTML report, either as a raw topology inventory or as a scorecard measured against a named target architecture. ## Core Features & Use Cases - Deterministic graph extraction: Scans workspace package.json files for runtime dependencies and peerDependencies, computing cycles (SCCs), fan-in/fan-out rankings, declared-tier violations, /api seam exposure, kit consumers, orphans, and per-package LOC. - Two report modes: A no-agent raw inventory mode that synthesizes the report spec automatically, and a target scorecard mode where the agent maps measured facts to target architecture invariants (holds/partial/open) with cited evidence. - Interactive HTML output: Renders a single self-contained HTML file with a D3 force/DAG graph (node area proportional to LOC, cycle edges in red, tier filtering, package and subpackage-circle views), Mermaid before/after diagrams, a verdict strip, scorecard table, finding cards, and a keystone recommendation. - Use Case: Run the topology launcher in a pnpm monorepo to get an instant dependency-graph audit showing cycles and layering violations, or name an architecture objective to produce a scorecard tracking migration distance versus architectural drift. ## Quick Start Run the bundled topology script from the workspace root to extract the dependency graph and open a rendered HTML topology report.
