style(docs): remove AI-writing tells from repository markdown

[kyle-sexton]

No linked issue

Summary

Repo-wide /ai-slop:audit fix pass over all 120 tracked markdown files. The detector found 1292 findings (1277 em dashes, 14 rule-of-three density hits, 1 emoji-formatting hit) across 86 files; this PR rewrites them per the plugin's rewrite guide while preserving what every sentence asserts.

Fix

• Em dashes in prose become colons (label bullets, heading qualifiers), commas, periods, or restructured sentences. Never parentheses, en dashes, or spaced hyphens.
• Rule-of-three density findings were judged individually; rhetorical triads collapse toward the strongest item, while load-bearing enumerations in normative criteria are kept and the reasons recorded.
• REVIEW.md's severity-marker emoji (load-bearing output vocabulary) are suppressed with an ai-slop ignore marker instead of removed.
• Work lands group by group (conventions/review first; components, conventions/engineering+process, docs, root/distribution follow) as parallel rewrite passes complete verification.

Verification

• Each group's rewrite is checked by a fresh-context semantic-diff reviewer, blind to the rewrite rationale, flagging any dropped qualifier, threshold, actor, or claim; flagged hunks are reverted. The conventions/review groups verified clean.
• Pre-commit lanes (typos, editorconfig-checker, gitleaks, markdownlint) pass on every commit.
• The ai-slop detector is re-run over the fixed set at the end; residual count reported in the PR thread.

Related

• melodic-software/ai-slop plugin (/ai-slop:audit), whose catalog and rewrite guide this pass applies.

🤖 Generated with Claude Code

https://claude.ai/code/session_01SdUrpi8EBT79LYEzDPHUP8

---

Generated by Claude Code

[claude]

Claude finished @kyle-sexton's task in 1m 40s —— View job

---

Claude is reviewing this PR

• Gather context (PR diff, changed files, task instructions)
• Run /review:code-review procedure (skill invocation didn't return content; proceeding with equivalent manual semantic-diff review per the fallback instructions)
• Analyze diff for correctness / maintainability issues (3 parallel semantic-diff reviewers dispatched across the 25 changed files)
• Post inline comments for anchorable findings
• Post summary review comment

[github-actions]

Claude has reviewed this PR 1 time. The lane skips further automatic reviews after 5; deleting this comment resets the count.

[kyle-sexton]

Cleanup complete; all five groups have landed. Detector re-run over the full 120-file tracked set:

• rule-em-dash: 1277 → 0. Two deliberate suppressions remain, each with a reasoned in-file ai-slop-ignore marker: the literal ISO 24495-1 title in conventions/engineering/concise-prose.md, and a verbatim quote of a stale manifest comment in docs/topics/standards-sync-audit/PLAN.md.
• rule-emoji-formatting: 1 → 0. REVIEW.md's 🔴/🟡/🟣 severity markers are load-bearing output vocabulary and are suppressed, not removed.
• rule-rule-of-three: 15 density findings remain by design. Every flagged triad was individually reviewed; the survivors are load-bearing enumerations in normative criteria (complete OS sets, WCAG terms, threat-model coverage lists). Four attempted collapses were reverted after a blind semantic-diff review flagged them as dropping list members.

Every rewritten group was checked by a fresh-context semantic-diff reviewer, blind to the rewrite rationale; the seven hunks flagged across all groups (narrowed open-class claims, dropped list members) were repaired or reverted. Pre-commit lanes (typos, editorconfig, gitleaks, markdownlint) pass on every commit, and the markdownlint fixture edit was verified against its component tests (24 assertions pass).

---

Generated by Claude Code