Sweep the semicolon prose class out of the carried docs

[ptr727]

The semicolon sweep only ever cleared GOVERNANCE.md. The other six carried files still ran the class, and they vendor downstream on the same no-downstream-trigger argument the dash sweep (#476) ran on.

The exemption came first

Sweeping the corpus straight away would have restructured the lists the exemption exists to protect, the same trap LABEL_DASH caught in #476. The rule keeps "a semicolon separating items in a list that already contains commas", but the check read the comma positionally (',' in prose[:m.start()]), so an enumeration whose commas fall in a later item had its openers flagged and its tail exempted, splitting one series in two. Three corrections, each with a test:

• The comma qualifies the list as a whole, not one separator's position.
• A markdown table row is judged one cell at a time. A row is a record of fields, and a comma in one column cannot excuse a semicolon in another.
• A bullet's **Label**: is dropped before the line is read, because it opens the bullet rather than announcing a list, the same construct the label dash is already exempted for.

Net on the carried corpus: 64 findings -> 60. Five lines went exempt (the D2/D4 guarantee lists and the S1 trace row, all genuine comma-carrying series), and two were newly caught.

The sweep

44 genuine clause joins recast as a comma or two sentences: CODESTYLE.md 24, .github/copilot-instructions.md 13, WORKFLOW.md 12, AUDIT.md 10, repo-config/README.md 1.

The 5A guarantees and the 5B trace table keep their semicolons, which is the standard series punctuation the rule already allows. Where a 5B cell carries no commas of its own (S2-S6, S10) it takes commas, so the table now punctuates each cell by what that cell contains.

Two splices that the relaxed exemption no longer flags (CODESTYLE.md "Spelling", WORKFLOW.md D4.1) are fixed by hand anyway, since I read them while sweeping.

Where this stops

All six files report zero, the bar GOVERNANCE.md already meets. The paren-internal asides they still carry ((default; strict baseline)-shaped) are the same construction GOVERNANCE.md kept through its own sweep, so holding the other files to a stricter bar than the rules file itself would be incoherent. The known limit is documented in scripts/README.md: a colon early in a long line still excuses a splice later on it.

Verification

prose_lint 130 cases and the full 180-case suite pass, repo_gate.py, spec/validate.py, spec/audit.py --selftest, the blocking prose run (charset, dupword, spelling) clean tree-wide, markdownlint 0 issues over 39 files, editorconfig-checker clean. cspell's gated scope (README + HISTORY) is clean, and the tree-wide 38-issue backlog is byte-identical to develop.

No downstream trigger, and this adds to the existing re-vendor debt rather than creating a new class of it.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR expands the “semicolon” prose sweep beyond GOVERNANCE.md by (1) updating the prose-lint rule’s exemption logic to match the documented intent more precisely (list-level comma detection, table-cell scoping, and bullet-label handling) and (2) applying the resulting sweep across the remaining carried documentation files so they reach zero findings consistently.

Changes:

• Refines semicolon detection/exemptions in scripts/prose_lint.py (list-level comma qualification, per-table-cell evaluation, and stripping - **Label**: bullet openers before analysis).
• Adds targeted regression tests covering the corrected semicolon exemption behaviors.
• Rewrites remaining carried-doc semicolon splices into comma/two-sentence forms while preserving allowed list-series semicolons.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
WORKFLOW.md	Recasts semicolon splices while preserving the workflow model and guarantee text intent.
scripts/test_prose_lint.py	Adds regression tests for list-level commas, per-table-cell checks, and bullet-label colon handling.
scripts/README.md	Documents the refined semicolon-rule interpretation and its known heuristic limit.
scripts/prose_lint.py	Implements span-based semicolon evaluation (table cells + label-colon stripping) and list-level comma exemption.
repo-config/README.md	Removes semicolon splice from carried prose without changing described ruleset behavior.
CODESTYLE.md	Sweeps semicolon splices throughout carried style guidance while keeping requirements intact.
AUDIT.md	Removes semicolon splices in carried audit procedure prose without changing audit semantics.
.github/copilot-instructions.md	Sweeps semicolon splices in carried Copilot/runbook instructions while preserving operational guidance.