Package Diagnostics Briefing
This chapter separates dependency shape, complexity pressure, and main-sequence distance into explicit diagnostic rules. Start with strict outliers where they exist; a Highest values group is a neutral fallback for orientation, not an AtlasArc problem verdict.
Subsystem-Level Architecture
Subsystem figures summarize boundary pressure, internal cohesion, and recursive architecture signals.
Where does boundary pressure concentrate across the visible architecture?
- How to read this graph: each bubble is a visible architecture-boundary rollup. Right means more outgoing boundary dependencies; up means more incoming boundary dependencies; size is total class count.
- Colour follows instability, so red means outgoing-heavy rather than automatically wrong. Use the figure to find boundary surfaces worth inspecting in Topology or Matrix.
- Dashed rings mark expanded aggregate boundaries whose visible descendants are also present. Terminal packages or source folders remain valid boundaries and use their own metrics.
Which subsystem areas contain internal cycle pressure?
- How to read this graph: each bubble is a visible architecture-boundary rollup. Right means more fully contained package-cycle groups; up means the largest contained package cycle is larger; size is package count.
- Only package cycles fully inside the boundary count here. Terminal-package class/file tangles remain separate Package Internals evidence in Topology. Use the cycle tables, Topology, or Matrix for concrete dependency edges and cross-boundary cycle context.
- Dashed rings mark expanded aggregate boundaries whose visible descendants are also present. Terminal packages or source folders remain valid boundaries and use their own metrics.
Source Folder Hotspots
Hotspot figures compare source folders through the available coupling and complexity evidence.
Which leaf source folders stand out by coupling pressure?
- Labels are shown for the largest report bubbles; inspect the source folder metrics table for the full ranked list.
Source Folder Diagnostic Signals
Use these short lists to find unusual dependency shape, complexity pressure, and main-sequence distance without treating every high value as a problem. Each rule is evaluated across the prepared report scope; cycle participation is shown only as context.
Dependency hubs
Which packages carry unusually broad dependency traffic in both directions?
These packages combine a wide consumer blast radius with broad outward knowledge. Understand their role before changing a contract or moving responsibilities.
Nothing in this report scope is separated enough from its peers to count as a statistical outlier. For orientation, the table shows 2 of the strongest eligible examples.
Continue in AtlasArcShow me the backboneIsolate the packages with the widest consumer blast radius and inspect who relies on them.How AtlasArc selected these rows
Rule: Fan-in or Fan-out is above its upper outlier cutoff and the other degree is at or above its 75th percentile.
Population: 31 real leaf architecture units with Fan-in and Fan-out. Result: No strict outlier result; showing 2 of 2 eligible fallback values.
Cutoff: Fan-in > 10.31 (31 eligible values); Fan-out > 7.67 (31 eligible values)
| Diagnostic evidence |
|---|
| Source Folderpackages.runtime-core.src |
|
| Source Folderpackages.runtime-dom.src |
|
Incoming-heavy packages
Which packages are unusually depended on while relying on relatively little?
These are load-bearing packages whose contracts may affect many callers even though their own outward dependency surface stays comparatively narrow.
AtlasArc found 1 value clearly separated from its peers in this report scope.
Continue in AtlasArcShow me the backboneTrace the callers that make these packages part of the project's backbone.How AtlasArc selected these rows
Rule: Fan-in is above its upper outlier cutoff, Instability is at most 0.35, and dependency hubs are excluded.
Population: 28 real leaf architecture units with Instability. Result: 1 qualified; showing 1.
Cutoff: Fan-in > 10.31 (31 eligible values)
| Diagnostic evidence |
|---|
| Source Folderpackages.shared.src |
|
Outgoing-heavy packages
Which packages depend unusually broadly without serving many callers?
These packages know about many parts of the system. They may be healthy coordinators, or responsibilities may have accumulated in one place.
Nothing in this report scope is separated enough from its peers to count as a statistical outlier. For orientation, the table shows 5 of the strongest eligible examples.
Continue in AtlasArcTicking time bombsInspect the outward dependency seams and decide whether the package's responsibility is still coherent.How AtlasArc selected these rows
Rule: Fan-out is above its upper outlier cutoff, Instability is at least 0.65, and dependency hubs are excluded.
Population: 28 real leaf architecture units with Instability. Result: No strict outlier result; showing 5 of 14 eligible fallback values.
Cutoff: Fan-out > 7.67 (31 eligible values)
| Diagnostic evidence |
|---|
| Source Folderpackages.compiler-sfc.src |
|
| Source Folderpackages.runtime-dom.src.modules |
|
| Source Folderpackages.compiler-sfc.src.script |
|
| Source Folderpackages.vue-compat.src |
|
| Source Folderpackages.runtime-core.src.compat |
|