Summary

extractAddedBlocks in review-enrichment/src/analyzers/duplication-scan.ts parses a unified diff into blocks of added significant lines with their new-file line numbers (used to locate copy-paste duplication). Inside the hunk loop it had:

if (!inHunk) continue;
if (line.startsWith("+++")) continue; // file header inside the patch, not an added line
if (line.startsWith("+")) { … newLine++; }

The +++ skip is meant to drop the +++ b/file diff header — but that header only appears in the patch preamble, before the first @@, which the if (!inHunk) continue; guard right above already drops. So inside a hunk, a line starting with +++ is never a header: it is an added line whose content begins with ++ (e.g. a pre-increment statement ++counterValueForLoop; → patch line +++counterValueForLoop;).

The stray skip therefore (a) drops that real added line from the block, and (b) since it continues before newLine++, shifts every following added line's number down by one. Downstream, scanDuplication → longestSharedRun reports DuplicationFinding.line / sourceLine off by one for any run following such a line, and a copy-pasted run that starts with a ++-line can be missed.

Fix: remove the redundant +++ skip so +++… lines fall through to the normal + handling (normalized and counted). Headers remain handled by the !inHunk guard.

Note: this is the only analyzer where the +++ skip is redundant — the others (secret-scan, redos, etc.) don't track inHunk and use the marker to skip preamble headers, so they have a different structure and are out of scope here.

No linked issue: small, self-contained diff-parsing fix in one helper; per this repo's linkedIssuePolicy: preferred, a direct PR with this rationale is appropriate.

Scope

• The PR title follows type(scope): short summary Conventional Commit format.
• This PR is focused and does not mix unrelated backend, UI, MCP, docs, dependency, and deploy changes.
• This follows CONTRIBUTING.md and does not reintroduce GitHub Pages, VitePress, site/, or CNAME.
• I linked an issue, or this is small enough that the summary explains why an issue is not needed.

Validation

• git diff --check
• npm --prefix review-enrichment run build (tsc clean)
• npm --prefix review-enrichment run test — the duplication-scan suite passes (36 tests) with a new regression test asserting an added ++counterValueForLoop; line is kept and the block's line numbers stay [10, 11, 12] (not [10, 11]). The test was confirmed to FAIL against the unpatched code (the line was dropped and numbers shifted).
• New/changed behavior has a regression test

If any required check was skipped, explain why:

• The change is one branch in a pure helper in review-enrichment/src; it adds/changes no analyzer, so analyzer-metadata.json is unaffected. Codecov's src/** patch gate does not apply to review-enrichment/**; this package's own node --test suite covers it.

Safety

• No secrets, wallet details, hotkeys, coldkeys, user PATs, private keys, raw trust scores, private rankings, or private maintainer evidence are exposed.
• Public GitHub text stays sanitized, low-noise, and does not imply compensation guarantees or optimization tactics.
• Auth, cookie, CORS, GitHub App, Cloudflare, or session changes include negative-path tests. (N/A — no such changes.)
• API/OpenAPI/MCP behavior is updated and tested where needed. (Internal diff parsing only; no schema change.)
• No visible UI change in this PR.

Notes

• Reproduction: extractAddedBlocks("@@ -1,0 +10,3 @@\n+++counterValueForLoop;\n+const secondSignificantLineHere = makeValue(2)\n+const thirdSignificantLineHere = makeValue(3)") previously returned one block with lineNos [10, 11] (the ++ line dropped, following lines shifted); it now returns lineNos [10, 11, 12] with all three lines.