[JarryShaw]

Please follow the guide below

• You will be asked some questions, please read them carefully and answer honestly

• Put an x into all the boxes [ ] relevant to your pull request (like that [x])

• Use Preview tab to see how your pull request will actually look like

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort)

• make test passes, and a test case covers the change — two new tests cover it; not the full make test, but the 12 related tests/project/tests/vendor modules (358 tests) pass

• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible — N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

Tick the commit type your subject line carries.

• fix — corrects a defect
• feat — adds a feature
• perf — changes performance, not behaviour
• refactor — changes neither behaviour nor performance
• test — tests only
• docs — documentation only
• ci — workflows or build tooling
• release — bumps the version or rolls up a distribution
• chore — anything else

---

Description of your pull request and other information

Commit changes in cron-vendor.yml ran git commit -am, which never stages an untracked file. verify-changed-files@v17 does report one (the ?? lines of git status -u --porcelain, per its entrypoint.sh). So a module generated for a new registry got a bump but was left out of the commit, and every later crawl bumped for it again.

• Fix: the step runs git add -A -- 'pcapkit/const/*.py' before git commit -am. That stages new, modified and deleted modules, and -a still picks up the files the bump rewrites. The pattern is quoted so git matches it at any depth; unquoted, the shell would expand it to pcapkit/const/__init__.py alone. It is limited to .py so a .<module>.py.XXXX.bak that _snapshot_and_restore could not remove is not committed: .gitignore does not cover those.
• Commit documentation changes (ci(vendor): commit a comment-only regeneration without bumping the version (#1674) #1684): no gap there. The classifier calls a new or deleted file code, so that run takes the bump path. A test walks the steps to show this, and the step keeps -a.
• Tests: tests/project/test_vendor_commits_new_modules_1685.py builds on ci: Vendor Update bumps a beta on a regeneration that changes only comments #1674's harness. It runs the step's script in a scratch repo holding a new, a modified and a deleted module plus two stray files, and checks that exactly the modules and the bumped __init__.py land in the commit. A second test checks that once a new module is committed, the next crawl reports nothing. Both fail on origin/main and pass here. A mutation check confirms each test catches an unquoted pattern, a whole-directory add, a repo-wide add, and a missing add.
• Docs: releasing.rst and workflows.rst each gain a sentence. actionlint reports no new findings (its 4 shellcheck notes are already on main).

Closes #1685

[JarryShaw]

Verdict: GOOD TO GO at afbb2a88a. A Sonnet cross-review checked the Opus worker's change. I re-ran the 1685 and 1674 tests: 25 passed.

The fix is one line before Vendor Update's commit: git add -A -- 'pcapkit/const/*.py'.

• What it stages: a new or deleted const module now lands with its bump, and -a still takes the bump's own files.
• The quoting: git's pathspec then reaches modules at any depth, rather than the shell expanding the pattern.
• Why .py only: a .bak that pcapkit-vendor failed to remove stays out of the commit.
• ci(vendor): commit a comment-only regeneration without bumping the version (#1674) #1684's documentation commit needs no change, because a new or deleted file always classifies as code.

The reviewer built its own harness, with new modules at depths 2, 4 and 6, a modified one, a deleted one, and five stray files. Only the crawl's output was committed. The test's four mutations (unquoted pattern, a bare git add -A, the directory without *.py, and no add at all) each fail.

This merges cleanly with #1677.

[github-actions]

Coverage: 89.46% (unit tier, Python 3.14, afbb2a88a, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	3	20	3	95.16%
pcapkit/const	18858	954	2366	868	91.06%
pcapkit/corekit	2366	67	760	35	96.48%
pcapkit/dumpkit	275	3	100	3	98.40%
pcapkit/foundation	3033	152	1076	35	94.23%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17166	226	4580	190	98.07%
pcapkit/toolkit	1416	279	584	5	77.80%
pcapkit/utilities	494	4	146	4	98.75%
pcapkit/vendor	4417	2342	1010	157	43.38%

Per-file detail: the coverage-html artifact of this run.