[JarryShaw]

Summary

Round 9 of the whole-repo prose sweep (#1103) -- a scoped pass over exactly the 4 items left outstanding from round 8 (#2428), not a fresh whole-repo sweep.

• pcapkit/protocols/protocol.py: flipped the 11 remaining unqualified ProtocolBase.<member> cross-references to Protocol.<member>, display and target both -- the same mechanical flip docs: prefer Protocol over ProtocolBase in cross-references, per the owner's #1103 ruling #2420/docs: retarget Protocol/ProtocolBase cross-references to match the #1103 display rule #2422/docs: #1103 round 8 -- finish the Protocol/ProtocolBase cross-reference sweep, fix other prose drift #2428 already did for 37+ other sites. Re-verified the carve-out list against the current file: the 2 __init_subclass__ sites (Protocol genuinely overrides it) and register's Raises: clause (named by docs(protocols): register's Raises name Protocol in some layers, ProtocolBase in others #2316/fix(foundation): registrar error messages and type hints name the public class, not the Base the guard checks #1016, test-locked by tests/protocols/test_register_class_guard_unit.py) are untouched.
• pcapkit/protocols/transport/transport.py: same flip for the ProtocolBase._lookup_registry reference fix(reassembly): the deferred TCP analyser dispatches on AppType, not bare ports #2421 added after docs: retarget Protocol/ProtocolBase cross-references to match the #1103 display rule #2422's sweep had already run, so it was the one site left behind.
• CONTRIBUTING.md / .github/PULL_REQUEST_TEMPLATE.md: added const to the valid commit-type list and the PR template checkbox. docs/source/contributing/conventions/process.rst already documented const as a Type label; CONTRIBUTING.md and the template simply hadn't caught up. tests/project/test_pr_template_commit_types_1102_unit.py derives its expected set dynamically from CONTRIBUTING.md, so no test change was needed -- it goes green with these two edits alone. Also fixed tests/project/test_conventions_doc_claims.py's now-false "const has no tickbox" docstring example, caught in review. (const: as a scope is in history via fix(const): accept RFC 2113's Router Alert 0 and IPX's own socket 0 #503/fix(const):; as a commit type prefix it has no precedent on main today -- the owner's ruling to add it stands on its own, not on prior usage.)
• docs/source/conf.py's two "Sphinx 9.1" comments were re-verified against the pinned toolchain (Sphinx==9.2.0, per docs/requirements.txt) rather than assumed stale: instrumented each code path directly (temporarily, not committed) and confirmed autodoc still hands is_enum_member the plain int for an enum member, and still prefers a pass-through __new__ over __init__ for a class signature (FTP renders (*args, **kwargs) with the workaround disabled). Both comments are accurate at 9.2.0 -- left unchanged.

No executable code changed -- every .py edit is inside a docstring. Verified with an AST-structure diff (docstrings stripped) against origin/main for both touched files: identical.

Test plan

• Searched for similar pull requests
• make pylint/make mypy/make isort -- not run (prose-only change, no code touched)
• make test equivalent: ran tests/project/ in full on this branch and on a clean main checkout under the same toolchain. Failure/error sets are byte-for-byte identical (234 failed/11 errors, all pre-existing and environmental -- subprocess tests that can't resolve util/_actions in this sandbox -- 2010 passed either way). Also built docs with the pinned toolchain (Sphinx==9.2.0) on this branch and on main: both produce exactly the same 1 pre-existing, unrelated warning (TransType cross-reference ambiguity).
• Changelog entry: 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?

• docs -- documentation only

Refs #1103.

[JarryShaw]

Review: NEEDS CHANGES at 57349ef91 (Sonnet). One real finding, text-only:

tests/project/test_conventions_doc_claims.py:1319-1320 still says "const is a type-ish label with no tickbox, so this is a one-way check by design" — false now that this PR adds the tickbox. The test still passes (it's a comment, not an assertion), but the sentence needs rewording or dropping.

Non-blocking: the PR body's dd036a52d precedent for const: as a commit-type prefix only exists on a PR ref (pr/503), not on origin/main (which merged as fix(const): … labelled fix) — origin/main has 0 const-typed subjects today, so "those are the ones in use" isn't literally true yet. Worth softening the claim, not blocking since the owner already approved adding const regardless of prior usage. Template puts const after the catch-all chore — cosmetic, your call.

Everything else verified independently: diff scoped to exactly the 4 claimed files (conf.py genuinely untouched), protocol.py/transport.py flips complete and correct (14→3 remaining dotted references, all in documented carve-outs), register's Raises: clause and its guard test untouched, the PR-template test genuinely parses dynamically (reverting either const edit alone fails it), both Sphinx 9.1 comments re-verified accurate at 9.2.0 via isolated probes against a 9.1.0 control, AST guard clean, docs build HTML diff limited to the 12 expected retargeted anchors, no scope creep.

[github-actions]

Coverage: 94.44% (unit tier, Python 3.15, da480c468, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	2	95.16%
pcapkit/const	19609	133	2620	120	98.84%
pcapkit/corekit	2605	74	818	37	96.52%
pcapkit/dumpkit	282	3	102	3	98.44%
pcapkit/foundation	3064	24	1068	26	98.79%
pcapkit/interface	138	4	52	3	96.32%
pcapkit/protocols	17593	193	4678	179	98.32%
pcapkit/toolkit	924	36	334	10	95.55%
pcapkit/utilities	502	7	144	5	98.14%
pcapkit/vendor	4449	1854	1020	167	55.11%

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

[JarryShaw]

Review: GOOD TO GO at da480c468. Verified directly (main thread). The stale const-specific example is dropped from test_conventions_doc_claims.py's docstring, reworded generically — confirmed correct, test passes clean (34 passed/193 subtests). Diff scoped to exactly the expected 5 files.

Both non-blocking items left as-is — both were explicitly optional in the last round's verdict.

[JarryShaw]

Correction to my own GOOD TO GO comment: I said both non-blocking items were left as-is. That's wrong for one of them — the worker did soften the dd036a52d precedent claim in the PR body (now reads "no precedent on main today"). Only the checkbox ordering was genuinely left alone. Verdict stands (GOOD TO GO), independently re-confirmed by a second review pass since.