Truemend
AUDIT
Code Quality Audit
halyard-api
Prepared for Halyard Systems
6.6
ABOVE SHIP LINE
Strong architecture and typing; dragged down by testing (4.4) and error handling (4.9). Nearly half the 30 findings apply for free.
LIVE FIX · VENDORS.TSSafe · $0
// try/catch only rethrows
async function saveVendor(d) {
try {
return await api(d);
} catch (e) {
throw e; // ← flagged
}
}
47
FILES
512
FUNCTIONS
47%
COST $0
Executive Summary
6.6
47
FILES
512
FUNCTIONS
41%
AI-WRITTEN
Findings by severity · 30
2 critical7 high14 medium7 low
Fix economics47% cost $0
14106
$0 · applied for you
Reviewed · bounded scope
A human decides
Weakest 5 of 40 · look here first
3.8
4.2
4.4
4.5
4.6
Scorecard · 7 families · 40
Forty signals rolled into families. Tap a family to see its dimensions.
Test strategy3.8
Test quality4.4
Assertion density4.9
WEIGHTED OVERALL6.6/ 10
Findings · filter by fix cost
WHAT WE FOUND
A catch block that only rethrows: it swallows nothing and adds nothing.
WHY IT MATTERS
Dead error handling hides intent and trips linters; readers assume it does something.
RECOMMENDED FIX
Remove the try/catch; the error already propagates unchanged.
catch_noop: truecyclomatic: 4 → 2
✓ re-parse + lintconfidence 0.98
How it’s made
01
Local analysis, not guesswork
Every score comes from analysis running on your machine, no model in the detection loop. 40 dimensions, mechanically scored and weighted.
02
Every finding carries its fix cost
Three tiers, so you always know what clearing it takes.
PatchesMechanical fix · $0 · file still compiles
RefactorsReviewed change · bounded scope
RewritesToo big to apply automatically
03
Fixes are proven before they count
Safe fixes validated by re-parse + lint, or the original is restored. Your code never leaves the machine.
Point it at your repo.
Watch it come true.
Watch it come true.
$ npx truemend init
truemend.aicode, made true