depwire/depwire

What changed

This release corrects a measurement bug in the dead code scoring logic. The Orphans & Dead Code dimension previously counted local variables, function parameters, interface properties, and class internals as candidates for dead code, which inflated dead code ratios for codebases with internally-complex modules.

v1.6.1 only counts symbols satisfying all of:

• Exported (visible outside the defining file)
• Zero dependents (no other symbol references it)
• Not in an exclusion list (entry points, framework callbacks, lifecycle methods)

Important — Health scores from v1.6.1+ are not directly comparable to earlier versions

This is a measurement fix, not a feature change. Your code did not change. Our measurement of it did.

If you compare a score from v1.5.0 or v1.6.0 against a score from v1.6.1+, expect the v1.6.1+ score to be higher in most cases. This is because the previous methodology was over-counting dead code.

For comparing health over time, use only v1.6.1+ scores as your baseline.

Validation results

On Depwire's own codebase:

• Overall score: 62/D → 67/D (+5)
• Orphans & Dead Code dimension: 20/F → 78/C (+58)
• Dead symbols: 6,578 → 270 (-6,308)
• Dead ratio: 88.5% → 3.6%
• All other dimensions: unchanged (no side effects)

On honojs (real-world TypeScript project):

• Overall score: 43/F → 45/F (+2)
• Orphans & Dead Code dimension: 20/F → 66/D (+46)
• Parser behavior: unchanged (352 files, 6,444 symbols, 2,188 edges identical before/after)

Testing

All 23 MCP tools tested and verified working. All CLI commands tested. Parser regression check confirms no change in parsing behavior.

Install

npm install -g [email protected]

Supported languages (16)

TypeScript · JavaScript · Python · Go · Rust · C · C# · Java · C++ · Kotlin · PHP · Swift · Mojo · Ruby · Dart · R