Code duplication

Detect copy-pasted code blocks across src/ and web/src/ with jscpd. Test files are excluded so the metric reflects production duplication only.

npm run duplication        # print all clones and the overall duplication %
npm run duplication:gate   # same, but exits non-zero if duplication exceeds the threshold

Reading the output

Each clone block names the two file locations that match:

Clone found (typescript)
 - commands/state.ts [11:64 - 34:58]
   state-format.ts [4:57 - 24:115]

The summary table at the end shows the total picture:

┌────────────┬────────────────┬─────────────┬──────────────┬──────────────┬──────────────────┐
│ Format     │ Files analyzed │ Total lines │ Total tokens │ Clones found │ Duplicated lines  │
├────────────┼────────────────┼─────────────┼──────────────┼──────────────┼──────────────────┤
│ typescript │ 64             │ 5469        │ 42226        │ 12           │ 152 (2.78%)       │
└────────────┴────────────────┴─────────────┴──────────────┴──────────────┴──────────────────┘

The Duplicated lines % is the key number. Under 3% is acceptable for this codebase; the gate enforces that ceiling.

Regression gate

.jscpd.json sets threshold: 3. npm run duplication:gate exits non-zero if the duplication percentage exceeds it, blocking regressions in CI. Ratchet the threshold down toward 2% as clones are removed.

Configuration lives in .jscpd.json. The minimum clone size is 5 lines / 50 tokens — shorter matches are noise, not duplication.